Methods and systems for generating Pareto-optimal smart contracts in Pareto-optimal decentralized applications.
By developing optimal smart contracts based on blockchain and AI/ML technologies and using oracle solving algorithms to generate Pareto optimal incentive mechanisms, the principal-agent problem and oracle problem in smart contracts are solved, minimizing agency costs and friction costs, and ensuring Pareto optimality and information transparency of contract relationships.
Patent Information
- Application Number
- CN202480088978.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-12-31
- Filing Date
- 2024-12-27
- Publication Date
- 2026-11-10
AI Technical Summary
Existing smart contracts suffer from principal-agent and oracle problems in value exchange, leading to increased friction and agency costs and failing to achieve Pareto optimality. Furthermore, traditional contracts and conventional smart contracts cannot effectively address the issues of information asymmetry and trust dependence.
By developing optimal smart contracts (OSCs) based on blockchain and AI/ML technologies, and using the Oracle Solving Algorithm (OSCRA) to generate Pareto optimal incentive mechanisms, the contractual relationship between the provider and the receiver is restored to balance under conditions of information asymmetry. Through the combination of off-chain AI and on-chain reputation scoring, information transparency and the effectiveness of the incentive mechanism are ensured.
It minimizes agency and friction costs in value exchange, provides Pareto optimality across all industries, ensures privacy and transparency for the recipient, and generates sustainable and competitive contractual relationships.
Smart Images

Figure CN123076885A_ABST
Abstract
Description
[0001] Cross-reference to related applications
[0002] This application claims priority to U.S. Provisional Application No. 63 / 616,763, filed December 31, 2023, the entire disclosure of which is incorporated herein by reference. Background Technology
[0003] Value exchange is central to every human economic activity and collaboration. Sellers and receivers agree on contracts that stipulate the terms and conditions for a fair and balanced transfer of value in all directions. The seller produces a product and provides it to the receiver in exchange for settlement. If the product is an asset (e.g., delivery of goods or services), the receiver pays cash to the seller. If the product is a liability (e.g., a transfer of risk or debt), the seller pays cash to the receiver. However, almost all value exchanges are affected by the principal-agent problem: the seller (agent) knows more about the product than the receiver (principal). The seller can use their informational advantage to ensure settlement is favorable to them, resulting in a net loss of value for the receiver, known as agency costs (including fraud cases). Information asymmetry about the characteristics of the product (adverse selection) or information asymmetry about the seller's behavior (moral hazard) threatens the equilibrium of the contract. Lack of transparent information and incentives reduces market efficiency associated with these contracts and can sometimes lead to market failure.
[0004] Traditionally, most of these contracts take the form of conventional contracts and are not economically efficient because interpreting them would leave them incomplete, and their formation, negotiation, performance, enforceability, and adversarial effect would incur high frictional costs (including often prohibitive intermediary costs, especially in the legal industry).
[0005] On the other hand, smart contracts reduce the need for intermediaries and are entirely determined by code. The decentralized nature of blockchain technology enables these smart contracts to be efficient, self-executing, and censorship-free. The use of smart contracts reduces frictional costs in value exchange and makes it possible to design directly executable incentive mechanisms; however, Pareto optimal parameterization of these mechanisms, integration of real-world data, and solutions to the principal-agent problem remain unresolved.
[0006] Therefore, there is a need for improved methods for generating and facilitating smart contracts. Summary of the Invention
[0007] This summary is intended to provide a brief overview of various subjects that will be elaborated upon in detail in the specific embodiments described below. This summary is not intended to identify key or essential aspects of the claimed invention. Nor is it intended to serve as an aid in determining the scope of the claims.
[0008] In some aspects, the technology described herein relates to a method comprising: receiving a set of mechanisms at a processor of one of a plurality of nodes, wherein each mechanism is defined at least by a provider action and a feasible contract; using the processor, defining a receiver loss function based on the set of mechanisms; using the processor, determining the mechanism among the set of mechanisms that minimizes the receiver loss function; using the processor, generating a smart contract based on the mechanisms; using the processor, broadcasting the smart contract to the plurality of nodes via a network interface that communicates electronically with the processor and a computer network, thereby deploying the smart contract to a blockchain maintained across the plurality of nodes connected via a computer network; receiving at the processor a transaction address corresponding to the smart contract from the blockchain; and using the processor, storing the transaction address in a database located on an electronic storage device at the node and communicating electronically with the processor.
[0009] In some aspects, the technology described herein relates to a system comprising: a processor of one of a plurality of nodes; a network interface that electronically communicates with the processor and a computer network; and an electronic storage device that electronically communicates with the processor, the electronic storage device storing a database; wherein the processor is configured to: receive a set of mechanisms, each mechanism being defined at least by a provider action and a feasible contract; define a receiver loss function based on the set of mechanisms; determine the mechanism among the set of mechanisms that minimizes the receiver loss function; generate a smart contract based on the mechanism; deploy the smart contract to a blockchain maintained across the plurality of nodes connected via a computer network by broadcasting the smart contract to the plurality of nodes via the network interface; receive transaction addresses corresponding to the smart contract from the blockchain; and store the transaction addresses in the database of the electronic storage device.
[0010] In some aspects, the technology described herein relates to a tangible, non-transitory computer-readable medium having instructions on it that, when implemented by a processor of one of a plurality of nodes, cause the processor to perform a method comprising: receiving a set of mechanisms, wherein each mechanism is defined at least by a provider action and a feasible contract; defining a receiver loss function based on the set of mechanisms; determining the mechanism among the set of mechanisms that minimizes the receiver loss function; generating a smart contract based on the mechanisms; broadcasting the smart contract to the plurality of nodes via a network interface that communicates electronically with the processor and the computer network, thereby deploying the smart contract to a blockchain maintained across the plurality of nodes connected via the computer network; receiving a transaction address corresponding to the smart contract from the blockchain; and storing the transaction address in a database located on an electronic storage device at the node and communicating electronically with the processor. Attached Figure Description
[0011] For a more complete understanding of the nature and purpose of this disclosure, reference should be made to the following detailed description given in conjunction with the accompanying drawings, wherein:
[0012] Figure 1 Example methods according to one or more embodiments of this document are illustrated, which illustrate a model based on an optimal smart contract solving algorithm.
[0013] Figure 2 A system for generating and implementing optimal smart contracts (OSCs) according to one or more embodiments of this document is illustrated.
[0014] Figure 3 A graph illustrating a casting function according to one or more embodiments of this document is shown.
[0015] Figure 4 An operating environment according to one or more embodiments of the present invention is illustrated.
[0016] Figure 5 This is a schematic diagram of example components of a device according to one or more embodiments of this document.
[0017] Figure 6 This is a schematic diagram of example components of a device according to one or more embodiments of this document.
[0018] Figure 7 This is a flowchart illustrating an example method according to one or more embodiments of this document.
[0019] Figure 8 An artificial neural network (ANN) according to one implementation is illustrated.
[0020] Figure 9The nodes of a machine learning model according to one implementation are illustrated.
[0021] Figure 10 An example is given of a method for training a machine learning module's machine learning model according to one implementation.
[0022] Figure 11 An example is given of a method for analyzing input data using a machine learning module according to one implementation. Detailed Implementation
[0023] The implementation method disclosed herein addresses the principal-agent problem by proposing a novel solution for generating and establishing smart contracts.
[0024] Crypto assets can be transferred on fully decentralized blockchains without relying on trust. However, for on-chain value exchange of Real-World Assets (RWA) and Real-World Liabilities (RWL), blockchain technology itself is not well-suited to avoid reliance on trust due to the oracle problem. The oracle problem refers to the inherent inability of blockchains and smart contracts to access real-world data or "off-chain" data. Oracles can provide smart contracts with data and a reliable source of information.
[0025] However, smart contracts lack access to oracles specific to value exchanges, particularly data related to product characteristics and the actions of the provider. Furthermore, they must provide off-chain computational intelligence to mitigate the principal-agent problem. If this is not achieved, savings in frictional costs will no longer be sufficient to offset the increased agency costs. Without addressing the oracle problem for this specific data, the irreversible nature of these value exchanges exacerbates the principal-agent problem.
[0026] Contract theory refers to the study and modeling of the behavior of economic agents in contractual relationships under conditions of information asymmetry. Contract theory can provide tools for determining Pareto optimal contracts, including signaling and screening to limit adverse selection, and the design of Pareto optimal incentive mechanisms to mitigate moral hazard. It can help develop profitable, competitive, and sustainable new products with well-structured incentives. Most importantly, it enables the recipient to avoid incurring high agency costs. However, due to the technological limitations of traditional contracts and conventional smart contracts, contract theory has not been fully applied in practice and cannot solve the principal-agent problem in real-world value exchange.
[0027] The implementation described in this paper addresses these shortcomings and enables the industrial application of contract theory. The implementation allows the recipient to minimize agency and friction costs in value exchanges with the provider, potentially covering all industries. The implementation solves the oracle problem, thereby resolving the friction costs associated with the need for a trusted third party and the principal-agent problem.
[0028] The implementation method described in this paper uses blockchain and AI / ML technologies to develop a general framework for optimal smart contracts (OSC), providing an application solution for the principal-agent problem and its underlying oracle problem in value exchange between providers and receivers.
[0029] While protecting the privacy of both the receiver and the provider, the system provides the receiver with the maximum amount of information about product characteristics and the provider's behavior. Next, an oracle-based OSC Solver Algorithm (OSCRA) is used to generate a Pareto-optimal incentive mechanism between the receiver and the provider, thereby restoring the balance of the contractual relationship in the event of initial information asymmetry.
[0030] One implementation approach is to achieve Pareto optimality across all value exchanges across all activity domains by customizing decentralized applications (DApps) for each corresponding industry. Governance of such DApps can be fully decentralized, for example, using decentralized autonomous organizations (DAOs), based solely on usage and entirely controlled by the recipients as DAO members, whose human attributes and uniqueness are verified through zero-knowledge proof protocols. For instance, the supply of an DAO's economic token can be slowly minted over time and with usage until it is used globally, ensuring the sustainable development of such DApps. Demand can be driven by usage.
[0031] The implementation can take into account provider type, provider actions or efforts, signals, and rewards as inputs to OSCRA. (Set of types) A collection of actions A collection of signals and the set of rewards It can be multidimensional, and even a wide-ranging function space. These sets and The σ-algebra of its measurable subsets can be used separately. and Related. Therefore, , , and It is a measurable space.
[0032] The set of feasible rewards can depend on the signal and be determined by... It means that it is The measurable subset. For each signal s, the set of feasible rewards given signal s is... .
[0033] Let △ denote a function that, given a measurable space... In the case of a set X, the function gives Let be the set of probability measures ℙ on a measurable subset of . .therefore, and Let be the probability space.
[0034] The provider's von Neumann-Morgenstein utility function can be expressed as: The receiver's von Neumann-Morgenstein loss (negative utility) function is expressed as follows: As shown below: and . and It is measurable.
[0035] Can Represented as an event Conditional probability measure for conditions and will Represented as an event Conditional probability measure.
[0036] The principle behind application display:
[0037] Definition 1. A mechanism is defined as any tuple. ,in and It is the transition probability, which satisfies the following condition for each... .
[0038] Defined as The mechanism can operate as follows: the type of provider can be determined by the state of nature. Extract from Θ. Once the provider knows its true type θ, it can report type θ' to the mechanism. The mechanism can then recommend action a' to the provider, which can be determined by a probability measure. Generation. After receiving the recommended action a', the provider can select action a from A. It should be noted that the provider's reporting of types and selection of actions do not need to change the fact that the contract can still remain "self-executing".
[0039] Finally, given by In the case of the generated signal s, the mechanism can be based on the probability measure The provider's reward, r, is generated. The signal can be generated based on the provider's true type and actions, while the reward can depend on the reported type and the recommended actions.
[0040] function For a given set of types and actions Give The conditional probability measure in this context is therefore interpreted as a contract. The formulation of the principal-agent problem in this paper is general enough to allow the receiver to randomize the reward offered to the provider based on observed signals. The provider can know the design of the contract before choosing its action.
[0041] Definition 2. Express the expected utility of a single provider as :
[0042]
[0043] Definition 3. Express the receiver's expected loss as :
[0044]
[0045] Definition 4. In the single-provider model, if for - Almost every type θ and every type θ' of a single provider will satisfy the incentive compatibility (IC) mechanism. Defined as:
[0046]
[0047] Will Indicated as in The set of mechanisms that satisfy incentive compatibility (IC) in a single-provider model (also known as incentive compatibility mechanisms).
[0048] Definition 5. If there exists a real number The mechanism that satisfies individual rationality (IR) will be met if the following conditions are met. Defined as:
[0049]
[0050] M IR Indicated as in The set of mechanisms that satisfy individual rationality (IR).
[0051] Definition 6. If mechanism The mechanism is defined as satisfying other constraints (OC) if it verifies any other constraints besides the incentive compatibility (IC) or individual rationality (IR) of the provider or receiver, such as constraints related to solvency, regulation, technology, market dynamics, etc.
[0052] M OC Represented as The set of mechanisms that satisfy other constraints (OC):
[0053] Definition 7. If mechanism If incentive compatibility (IC), individual rationality (IR), and other constraints (OC) are satisfied, then the mechanism is defined as a feasible mechanism.
[0054] Will Represented as A set of feasible mechanisms:
[0055] For any feasible mechanism ,make
[0056]
[0057] This represents the recipient's expected loss if the provider reports truthfully and takes the recommended actions.
[0058] Definition 8. For the set of feasible mechanisms The principal-agent problem between the provider (agent) and the recipient (principal) is as follows:
[0059] .
[0060] The set of solutions to this principal-agent problem is called the Pareto optimal mechanism, and thus limits the Pareto optimal contract solutions associated with the provider and receiver.
[0061] According to Kadan, Reny, and Swinkels (2017), there exists a set of known conditions that can be used to prove the Pareto optimal mechanism and thus prove the existence of Pareto optimal contract solutions.
[0062] Assumption
[0063] This paper extensively cites the definition and main results of the existence of Pareto optimal mechanisms in the paper "Existence of optimal mechanisms in principal-agent problems" (2017) by Kadan, Reny, and Swinkels (hereinafter referred to as KRS), the entire contents of which are incorporated herein by reference.
[0064] To recap, Polish space is a separable topological space that can be measured by a complete metric. Any Polish space... The measurable subset will always be the Borel set. ,and This will represent a topology equipped with weak convergence (weak * topology). The space of probability measures.
[0065] Environmental Space Assumptions
[0066] Assumption 1. Let Θ be a non-empty Polish notation space, and It is firm.
[0067] Assume 2. Φ is a closed subset, where for each The set is non-empty.
[0068] Benefit assumptions
[0069] Assumption 3. and It is measurable and has a lower bound, which is bounded to 0 without loss of generality; and, for each as well as - in almost every case , exist For all It is the lower half of the continuous sequence.
[0070] Assumption 4. For any type θ, any as well as any compact subset , and The closure is tight.
[0071] Assumption 5. For any type θ and any sequence in ,if ,but .
[0072] Information assumptions
[0073] Assumption 6. Transition Probability So that for each θ, The relationship between 'a' and 'a' is continuous.
[0074] Assumption 7. Exists family of subsets This makes it possible for all , yes a measurable subset; and for all And all measurable All have .
[0075] According to hypothesis 7, when both measures are restricted to When the measurable subset, Compared to It is absolutely continuous. Equivalently, the unrestricted measure on a measurable subset of S. Compared to unrestricted measure It is absolutely continuous. Applying the Radon-Nicotinim theorem to unrestricted measures, there exists a Radon-Nicotinim derivative defined over the entire S, and therefore... express Compared to Any version of the Radon-Nicodim derivative.
[0076] Assumption 8. For all and each sequence ,exist And the Radon-Nicodim derivative exists. and The version that makes
[0077]
[0078] Assumption 9. If the family of sets in Assumption 7 Makes for a certain have Then it exists This makes it possible for all have And for all have .
[0079] For many applications, all nine hypotheses can be verified by restating the model, thus requiring only the testing of hypotheses 1 through 5. A primary case is that all spaces are Euclidean spaces, and the signal distribution has a property of... The density is continuous with θ.
[0080] theorem
[0081] Theorem 1. In The above bestows -Measurement. Under assumptions 1 to 9, for all c ≥ 0, It is a compact subset of M, and the client's loss function is defined in (1). It is the lower half of the continuous sequence. Therefore, if If it is not empty, then the client's problem (2) has a solution.
[0082] We can now cite the main results from Kadan, Reny, and Swinkels.
[0083] Theorem 2. If Assumptions 1 to 9 are satisfied, then the principal-agent problem has a solution as long as at least one incentive-compatible mechanism exists.
[0084] The set of incentive-compatible mechanisms can be quantified such that: (i) for any constant, a subset of incentive-compatible mechanisms that result in the principal's expected loss being weakly below that constant is compact; and (ii) the principal's loss function as a function of the mechanism is lower-semi-continuous, and the set of incentive-compatible mechanisms is non-empty. Although the existence of at least one incentive-compatible mechanism is usually readily determined in applications, the following proposition (Proposition 1) provides a general condition for the existence of incentive-compatible mechanisms.
[0085] Proposition 1. Assume that 1, 3, and 6 are true, and that there exist measurable... So that for each s, there is And make
[0086]
[0087] It is continuous for each θ.
[0088] There exists an incentive-compatible mechanism, that is... Not empty.
[0089] The continuity requirement is satisfied if, assuming all other assumptions hold, either (a) a fixed reward is always feasible and the agent's utility in obtaining that reward is independent of the signal; or (b) the function It is continuous, and for each ,about They are consistent.
[0090] Theorem 2 is derived directly from Theorem 1. According to Theorem 1, the set of incentive-compatible mechanisms can be quantified such that: (i) for any constant, the subset of incentive-compatible mechanisms that cause the principal's expected loss to be weakly lower than that constant is compact; and (ii) the principal's loss function as a function of the mechanism is lower semi-continuous, and the set of incentive-compatible mechanisms is non-empty. Although it is usually easy to determine the existence of at least one incentive-compatible mechanism in applications, Proposition 1 gives the general condition for the existence of incentive-compatible mechanisms.
[0091] Multi-provider principal-agent model
[0092] Now introduce For the index provider and The receiver is the indexed party. Without loss of generality, in the context of this section, since we are discussing receiver j, j is omitted.
[0093] In this section, KRS shows how Theorem 1 derives the existence result for a model of interaction between a single receiver and multiple providers. This is achieved by adding a subscript to each of the I providers. Otherwise, the symbols will be the same as in the single-provider case. Therefore, and It is the type space and action space of provider i, and defines... and Assuming the provider's type is based on from Extraction from the middle. The utility function of provider i is... And the receiver's loss function is The mechanism here is... Its specified transition probability and And for all have .
[0094] mechanism It operates as follows. The natural state is based on... Extract the type vector θ. Each provider i simultaneously learns its type. And then privately report its type to the mechanism. For any vector of reported types The mechanism is based on Vector of choice of action Then, each provider i was privately informed of the action they recommended. Subsequently, all providers simultaneously select an action, where provider i chooses from... Select an action. If the vector of the selected action is... According to Extract signal Finally, according to Extracting Remuneration Provider i obtains utility The recipient suffers the loss. .
[0095] Absolute Continuity: Type and Signal
[0096] For each provider i, let express exist The edge distribution on, and let This represents the product of these marginal distributions. From this assumption... The product relative to its marginal distribution It is absolutely continuous. Therefore, according to the Radon-Nicotim theorem, there exists a measurable... This makes it possible for each measurable ,
[0097]
[0098] Assumption 10. Exists This makes it possible for each , Compared to It is absolutely continuous. Specifically, it assumes the existence of measurable... This makes it possible for each as well as Each measurable subset ,
[0099]
[0100] From this point on, we also assume that assumption 10 holds true. This is the same assumption made in the single-provider model to obtain the "standard case".
[0101] Incentive compatibility
[0102] For any mechanism Given any provider i, according to Proposition 7.27 of Bertsekas and Shreve (1978), it can be... Decomposed into , and For some transition probabilities. Furthermore, the transition probabilities... remove The set of types with measure zero is unique outside of it; and the transition probability remove It is unique outside a certain subset, and that subset is based on exist The marginal distribution on has zero measure. It should be noted that this marginal distribution is equal to .because Only Since it is almost always unique, the most natural approach is to define incentive compatibility as the fact that only almost (but not all) false reports do not improve the outcome.
[0103] Definition 9. In a multi-provider model, if for each provider i, there exists a measurable... Make And for all The incentive compatibility (IC) mechanism will be satisfied if the following conditions are met. Defined as:
[0104]
[0105] Indicates the following The set of mechanisms that satisfy incentive compatibility (IC) in a multi-provider model (also known as incentive compatibility mechanisms):
[0106] Definition 10. In In the case of a multi-provider model, the expected utility of provider i is... Represented as:
[0107]
[0108] In order to obtain by type Conditional expected utility, when When it is positive, divide both sides of the equation in Definition 9 by... .
[0109] about The condition is of great significance: certain conditional distributions are obtained only through the Radon-Nicodim derivative. Depends on the true type of i These conditional distributions first occur when i is of type i. And it reported In this case, the recommended action The conditional distribution; and secondly, in i of type And its report type Then asked to take action In this case, the actions and types of other providers The conditional distribution. This is the conditional distribution in Definition 9. and It depends only on the type reported. And not depending on the real type The reason.
[0110] In transition probability or Changing these transition probabilities on any non-uniquely defined set of zero measure will not affect Does it satisfy Definition 9? Therefore, any The incentive compatibility does not depend on the chosen transition probability. and The specific version.
[0111] Exported single provider i model
[0112] This section shows that, from the perspective of each provider i, the multi-provider model can be simplified to a principal-agent model involving only that single provider i. The idea is to interpret all aspects of the problem involving other providers as part of the reward space of provider i, and to eliminate any correlation between types and any dependency of signals on actions or types by rescaling the payoff of provider i using the density of types and signals.
[0113] The key property of the resulting "single provider i" model is that a mechanism is incentive-compatible in the multi-provider model if and only if the mechanism derived for each provider i is incentive-compatible in the single provider i model (Proposition 2). KRS establishes an existence result for the multi-provider model based on the result established for the single provider model (Theorem 3).
[0114] Given the above multi-provider model, for any provider The single-provider i model is defined as a principal-agent model with a single provider, but the reward space is... The signal space is S, and the signal always depends on... Extract, action space is The type space is a priori of And for any The utility of provider i is And the recipient's loss is Finally, for any , The set of feasible returns in the middle is .
[0115] Therefore, the mechanism of the single-provider i model is a pair of transition probabilities. , making And for each have .
[0116] The proof of the next proposition (Proposition 2) is in the appendix of the KRS, which connects the single-provider i model with the multi-provider model.
[0117] Proposition 2. A transition probability exists for every provider i if and only if there is one. and , making and When the single-provider i model is incentive-compatible, the mechanism The incentive is compatible for the multi-provider model.
[0118] The proof of the next result is in the appendix of KRS. This proof depends on Proposition 2 and Theorem 1.
[0119] Theorem 3. Assume that for each provider The single-provider i model satisfies assumptions 1 to 5. Therefore, as long as there is at least one incentive-compatible mechanism among multiple providers, there exists a mechanism in the multiple-provider model that minimizes the expected loss of the receiver.
[0120] The conditions for the existence of at least one multi-provider incentive compatible mechanism are as follows.
[0121] Proposition 3. Assume that 1, 3, and 6 are true, and that there exists a measurable function. and So that for each s, there is , ; and make Assuming for each provider ,and For each If all values are continuous with respect to a, then there exists at least one multi-provider incentive-compatible mechanism.
[0122] OSCRA algorithm
[0123] This can be used to determine the OSC (Optical Constraints). An OSC can include a hybrid smart contract that solves the principal-agent problem and its underlying oracle problem in value exchange between a provider and a receiver. The OSC can be built on blockchain, AI / ML technologies, and contract theory. The oracle can include a combination of off-chain AI and on-chain reputation scoring / certificates, based on privacy-enhancing techniques that guarantee strict privacy for both the receiver and the provider. The OSC can achieve Pareto optimality by solving the OSCRA (Optical Constraint Argument). Therefore, the OSC can generate Pareto optimal incentives between the provider and the receiver, thereby restoring balance to the contractual relationship in the event of initial information asymmetry. The OSCRA can include modeling the provider's effort and effort cost, modeling the utility functions of the receiver and the provider, and estimating the value of the preserved utility (e.g., the provider's suboptimal choice in a competitive market). Thus, an OSC can be a smart contract that, under constraints, minimizes the receiver's expected loss (e.g., maximizes the receiver's expected utility) as a function of the provider's effort and the receiver's feasible contract parameters. Depending on the use case, this constraint can include:
[0124] • Incentive compatibility (IC): The provider can choose the contract that maximizes its expected utility function from all feasible contracts that the receiver can agree to;
[0125] • Individual rationality (IR): A provider may accept a contract only if the utility generated by the effort exceeds the provider's retained utility; or
[0126] Solvency constraint (SC): All components of the provider's capital must remain positive (e.g., financial capital and health capital).
[0127] These constraints are generally applicable to free market scenarios. Other contracts may be subject to other constraints (e.g., regulatory, technological, market dynamics), which must be considered in a specific way for each industry. For example, these constraints can be considered in the design and modeling of probability spaces for type, action, signal, and reward, as well as the utility function of the provider and the loss function of the receiver.
[0128] Optimal smart contracts can include hybrid smart contracts that address both the principal-agent problem and the underlying oracle problem in value exchange between a provider and a receiver within a given industry. Industry-specific oracles based on AI and blockchain technologies can combine information about the value exchange to consistently achieve Pareto optimality between the provider and receiver.
[0129] In this context, in order to solve the generalized principal-agent model described in the previous section, oracles are required to provide reliable real-world information about products of complex nature (such as goods, services, or liabilities), as well as utility functions for both the provider and the receiver.
[0130] An optimal smart contract can be a hybrid smart contract that, under constraints, minimizes the receiver's expected loss (e.g., maximizes the receiver's expected utility) based on the provider's effort and the parameters of the receiver's feasible contracts. Depending on the industry and specific use case, this constraint typically looks like this:
[0131] • Incentive compatibility (IC): The provider selects the contract that maximizes its expected utility function from all feasible contracts that the receiver can agree to;
[0132] • Individual rationality (IR): A provider accepts a contract only if the utility generated by the effort exceeds the provider's retained utility; and
[0133] Solvency constraint (SC): All components of the provider's capital must remain positive (e.g., financial capital and health capital).
[0134] These constraints are generally applicable to free market scenarios. Other contracts may be subject to other constraints (regulation, technology, market dynamics), which must be considered in a specific way for each industry. The framework proposed by Kadan, Reny, and Swinkels (2017) has a very broad formulation that allows all these constraints to be considered, especially when designing and modeling probability spaces for types, actions, signals, and rewards, as well as utility functions for providers and loss functions for receivers.
[0135] The Optimal Smart Contract Solving Algorithm (OSCRA) can be an algorithm that solves the generalized principal-agent model between the provider and the receiver based on product information in a given industry and the utility functions of both the provider and the receiver, and generates a set of optimal smart contracts that can be signed by both the provider and the receiver.
[0136] Optimal smart contracts achieve Pareto optimality by solving the Optimal Smart Contract Solving Algorithm (OSCRA). Therefore, optimal smart contracts generate Pareto optimal incentives between the provider and receiver, restoring balance to the contractual relationship even when initial information asymmetry exists. The main steps of the OSCRA algorithm are modeling the effort and effort cost of the provider, modeling the utility functions of both the receiver and the provider, and estimating the value of the reserve utility (the provider's suboptimal choice in a competitive market). These algorithms must be designed, calibrated, tested, and validated according to the characteristics and requirements of each use case. These algorithms are industry-specific and will be a subject of dedicated research each time.
[0137] Then, the parameters of the Pareto optimal contract solution of the principal-agent model studied in the theoretical part are encoded into a smart contract. This smart contract then binds the provider and receiver, ensuring transparency and trust between them regarding the protocol binding them. Therefore, optimal smart contracts can implement blockchain and AI technologies applied to contract theory.
[0138] OSCRA and the underlying oracles of optimal smart contracts
[0139] To solve the OSCRA algorithm and generate optimal smart contracts, off-chain data needs to be accessed to model all inputs and solve the Pareto optimality mechanism problem. This allows for the identification of the parameters of a Pareto optimal contract between the provider and receiver. A similar oracle problem arises in executing optimal smart contracts and consistently verifying consistency between the provider and receiver. Therefore, specialized oracles need to be developed for each industry and use case. Currently, neutral, impartial, and high-quality information about products in different markets is typically obtained through experts; however, acquiring this information is extremely costly and often unaffordable, exacerbating the information asymmetry problem associated with the principal-agent problem. Artificial intelligence (AI) technology can significantly reduce the cost of acquiring this information and appears to be a good candidate technology for addressing the underlying oracle problem of the principal-agent problem. The potential use of AI technology and other mechanisms enables the acquisition of knowledge about the transferred product (whether goods, services, risks, or liabilities) and the components of the utility functions of the provider and receiver. During each product transfer iteration, the provider's reputation and certificate ratings are calculated, updated, and stored on-chain to prevent tampering. Therefore, an oracle can be a combination of off-chain computation using AI and on-chain reputation / certificate scoring. The oracle should be based on privacy-enhancing technologies that ensure strict confidentiality for both the receiver and the provider. These desired characteristics of oracles will be industry-specific and will be a subject of dedicated research each time.
[0140] Pareto optimal DApps or optimal DApps can include a set of industry-specific decentralized applications (DApps) that use the OSCRA algorithm to generate optimal smart contracts between the provider and the receiver, enabling the provider and receiver to sign and commit to fulfilling a given optimal smart contract.
[0141] Optimal DApps can be implemented across various activity domains. Each optimal DApp can provide a marketplace, enabling recipients and providers to enter into optimal smart contracts. First, use case-specific systems are designed to maximize the transparency of provider characteristics and behavior. Then, oracles solve the OSCRA algorithm. By observing a fairly accurate approximation of the provider's actual effort, strong on-chain incentives are created, making the cost of reputational damage due to adverse behavior high. In addition to reputation, an on-chain fund retention mechanism at the start of the contract subsequently rewards the provider who exerts the most effort and thus penalizes (or reduces rewards) the provider who exerts the least effort. Oracles also help providers deliver better services / goods or reduce the frequency and intensity of risks.
[0142] Figure 1 Example method 100 according to one or more embodiments of this document is illustrated, which illustrates a model based on an optimal smart contract solving algorithm.
[0143] The aforementioned models include parametric function optimization models. This study used R open-source statistical analysis software version 3.4.0 (R Foundation for Statistical Computing, Vienna, Austria; http: / / www.R-project.org / ). This software provides:
[0144] • Management of a given database (data processing and analysis);
[0145] • Statistical analysis used to estimate probability distributions based on data;
[0146] • Used for simplifying formal computations (or symbolic computations) of complex expressions, including integral computations and optimization computations;
[0147] • Perform numerical analysis when an expression cannot be evaluated symbolically.
[0148] After solving the model, blockchain technology was used to encode the optimal smart contract.
[0149] The optimal smart contract solution algorithm involves determining contract features that minimize the receiver's expected loss (equivalent to maximizing expected utility) by maximizing the provider's expected utility. These contract features are most advantageous to the receiver, and under the conditions of this contract, the provider will reveal its type and choose the most suitable action, while also verifying that the contract remains sufficiently attractive to the provider. The corresponding optimal smart contract as the solution to the algorithm will then be given. To recap, the provider's expected utility in a single-provider model... The expression is (Definition 2):
[0150]
[0151] In a multi-provider model, when At that time, the expected utility of provider i For (Definition 10):
[0152]
[0153] In both scenarios, the recipient's expected loss ( The expression for ) is (Definition 3):
[0154]
[0155] Preparation of the OSCRA algorithm
[0156] At point 101, the process can begin, and at point 105, the algorithm can receive the input dataset.
[0157] At point 110, the optimal smart contract solution algorithm can be prepared, which includes steps 111 to 114:
[0158] First, at point 111, we can design the components of the principal-agent problem (optimal smart contract model):
[0159] • Design reward functions, regardless of whether the contract is deterministic (e.g., a function from the signal space to deterministic rewards) or randomized (e.g., a function from the signal space to a reward probability distribution).
[0160] • The structure of the probability space for design types, actions, signals (and rewards when the contract randomizes the rewards) (the complexity depends on the use case and available data; in other words, the sample space associated with these probability spaces can be a finite set, real numbers, vector space, function space, etc.).
[0161] • Design type Q, action A, signal S (and reward R when the contract randomizes the reward) random variables (which can be discrete random variables or real-valued random variables, random vectors, random processes, random functions or random fields, etc.), and parameterize them using type, action, signal and reward indices;
[0162] • Design the provider's utility function and parameterize it (including the effort cost function);
[0163] • Design the receiver's loss function;
[0164] • Design constraints for principal-agent problems (incentive compatibility, individual rationality, including retention utility design, solvency constraints, etc.).
[0165] Secondly, at point 112, data mining can be performed using available data while respecting the privacy of members and partners. Data mining involves selecting and ensuring the reliability of the raw data, then synthesizing the raw data and creating new metrics. Another part of the model input is generated using statistical methods called "data analysis," such as data dimensionality reduction. Predictive models typically assume linear independence: explanatory variables should not be collinear. The aim is to reduce the dimensionality of the problem.
[0166] • Input / data required for the integrated calibration outsourcing-proxy model;
[0167] • Homogeneous categories of potential types and action builders; and
[0168] • Segmented based on potential type and action execution.
[0169] Third, at point 113, an indicator can be constructed:
[0170] The type index measures the characteristics of a product (hidden information for the recipient). A low type index indicates a high probability of adverse selection in the recipient-supplier relationship. The higher the type index value, the greater the increase in the recipient's expected utility, and therefore the lower their expected loss.
[0171] • Action metrics measure the provider's behavior (hidden actions from the receiver's perspective). Low action metrics indicate a high probability of moral hazard in the receiver-provider relationship. The higher the action metric value, the greater the increase in the receiver's expected utility, and therefore the lower their expected loss. Depending on the use case, action metrics can randomly depend on type metrics.
[0172] • Signaling metrics measure the success (output / result) of a provider in fulfilling a contract. Signaling metrics typically depend randomly on type metrics and action metrics.
[0173] • Compensation metric, which measures the compensation a recipient pays to a service provider in exchange for contractual performance, and is therefore directly dependent on the signal. In the case of delivery of goods and services, this compensation is typically positive; in the case of an exchange that transfers responsibility (such as risk or liability) from the provider to the recipient, the compensation is negative (the provider makes one or more positive cash settlements to the recipient to settle the contract).
[0174] Fourth, at point 114, model calibration is provided by constructing indicators, which enables estimation of the probability distributions of random variables designed and parameterized during the design phase:
[0175] • Estimate the marginal probability distributions of the random variables for type Q, action A, signal S (and reward R when the contract randomizes the reward) for each segment. There is information asymmetry in the distributions of type and action: for the provider, these distributions are known and deterministic, while for the receiver, the type and action indicators are random variables. However, since the signal originates from type and action, there is no information asymmetry in the signal: in some cases, it can be assumed that the receiver's information about the signal random variables is consistent with the provider's information;
[0176] Estimate the joint probability distribution of the random variables of type Q, action A, signal S (and payoff R when the contract randomizes the payoff) on each segment. Associated (potentially parameterized) Copula functions;
[0177] • Estimate the joint probability distribution and conditional probability distribution , and .
[0178] • Calibrate other components of the principal-agent model (especially the provider's utility function and the receiver's loss function).
[0179] OSCRA algorithm initialization
[0180] Once the components of the model have been designed, parameterized, and implemented, the OSCRA algorithm can be initialized at point 115 for the new receiver (j) that becomes a member of the DAO.
[0181] Before using the ecosystem, the receiver (j) and the provider An initial contract was entered into. When a new member (j) joins, the characteristics of the provider can be learned.
[0182] Initial mechanism It is structurally feasible and satisfies incentive compatibility (IC), individual rationality (IR), and other constraints (OC) (Definition 7). In practice, this is achieved by signing an initial contract (e.g., executing a smart contract, signing a contract electronically, etc.).
[0183] • Provider The action that best suits its purpose under the initial contract conditions has been chosen. Therefore, the excitation compatibility (IC) in Definition 9 is verified.
[0184] • Provider This verifies the individual rationality (IR) in Definition 5, because it did not choose another contract or option (preserving utility), but instead chose the conditions of the initial contract; and
[0185] • Provider The recipient verified other constraints (OCs) related to solvency, regulation, technology, market dynamics, etc., as defined in Definition 6.
[0186] This in particular allows for the estimation of the following data:
[0187] • Provider Type Indicators and Actions Indicators and utility functions and retention of utility ;as well as
[0188] • Loss function of the receiver .
[0189] Algorithm initialization can include estimation (provided by the provider) in the initialization mechanism. initial type and action Utility function and retention of utility and the receiver's loss function ):
[0190] • Estimation type and action The initial value;
[0191] • Calibrate the initial contract The parameters.
[0192] According to the provider Risk aversion level calibrated utility function Parameters (including effort cost function) The provider's risk aversion and effort aversion are intrinsic characteristics, independent of the contract provided to the provider: the utility function and effort negative utility function are also the same as in the initial contract;
[0193] • Calibrate the receiver's loss function ;
[0194] • Calibrate the constraints of the principal-agent problem (incentive compatibility, individual rationality, including retention utility design, solvency constraints, etc.).
[0195] These estimates provide the algorithm with information for the next cycle of data preparation / recalibration phase / input dataset 105.
[0196] Once the algorithm has completed initialization, the parameters of the expected utility function and effort cost are calibrated, regardless of the contract between the provider and the receiver.
[0197] Solving the principal-agent problem
[0198] At point 120, the principal-agent problem (optimal smart contract model) defined in point 8 can be solved.
[0199] This is a constrained nonlinear optimization problem, in which solving the constraints (excitation constraint IC (Definition 9), participation constraint IR (Definition 5), and other constraints OC (Definition 6)) is itself a constrained nonlinear optimization problem.
[0200] The provider's plan is determined to meet the following conditions. :
[0201] .
[0202] The solution to the provider's plan is Above has A set of feasible mechanisms
[0203] The recipient's plan is determined to meet the following conditions. :
[0204] .
[0205] For example, the augmented Lagrangian method for provider i can be used to solve the receiver's plan. This corresponds to a reward structure that incentivizes provider i to reveal its true type and take the recommended action while minimizing the receiver's loss. The convergence of the optimization computation completes the model solution algorithm, yields the optimal smart contract, and the algorithm is then validated.
[0206] At point 125, the output dataset of the optimal smart contract model can be generated.
[0207] The parameters can then be encoded into the smart contract.
[0208] At point 130, it is possible to assess whether the provider is a new provider.
[0209] • If so, then at point 140, all the best smart contracts can be provided to the provider for selection.
[0210] Otherwise, if the provider already has a contract, it can evaluate at point 135 whether the current contract belongs to the set of optimal smart contracts. If so, it can wait at point 150 for the contract's renewal date and new data to restart the algorithm.
[0211] Alternatively, at point 140, all the best smart contracts can be offered to the provider for selection.
[0212] • At point 140, the recipient can be offered the option to order one of the best smart contracts among the best smart contracts.
[0213] If the recipient accepts and signs at point 145, they can wait at point 150 for the contract renewal date and new data to restart the algorithm.
[0214] Otherwise, the process ends at point 160.
[0215] A DAO is a member-owned community without centralized management, which can provide a structure for solving significant economic and organizational governance problems. Therefore, OSCRA can be delivered via a DAO configured to facilitate Pareto optimality in cross-industry value exchange. One or more DAOs can be used to implement DApps configured to generate the OSC disclosed herein.
[0216] One or more tokens can facilitate matters related to the DAO:
[0217] • Tokens can be used, which may include non-fungible tokens (NFTs) that are non-transferable and non-destructible. These tokens can be configured as usage metrics and grant specific rights to recipients who are DAO members. For example, a member could be allocated one such token for each full year of use of at least one DApp in the ecosystem. Therefore, the supply of this token can be unlimited and strictly non-decreasing.
[0218] • Board tokens, which may include transferable and non-destroyable NFT tokens allocated to members of the DAO board, are used to vote on matters related to the flexible execution of DAO operational decisions and the provision of protection in the event of incidents. Their supply can remain constant. They can initially be allocated to founders and early advisors. They can be transferred through governance votes, based on the reputation of the DAO's board members, and are designed to achieve an optimal trade-off between flexibility and decentralization for each stage of the project.
[0219] • Governance tokens, which can include non-transferable and burnable voting NFTs, can be used, for example, to vote on DAO strategic decisions. Each usage token associated with an active DAO member, as well as each board token, grants them one governance token. Therefore, at DAO launch, the supply of governance tokens can be equal to the supply of board tokens to allow for flexible DAO growth. Over time, as governance tokens are distributed through usage, the board's weight in DAO strategic decisions can become negligible compared to the weight of DAO members. This shift allows for reduced operational risk in the early stages of the project by scaling it up in a flexible but centralized manner, while achieving full decentralization in the long term.
[0220] • Economic tokens, which can be configured, for example, to be based on existing cryptocurrencies, such as Bitcoin.
[0221] Figure 2 A system 200 for generating and implementing OSC according to one or more embodiments of this document is illustrated.
[0222] Each DApp 210 associated with DAO 202 can provide a marketplace, enabling DAO members such as receiver 222 and DAO partners such as provider 224 to agree on OSC terms such as OSC 226. A use case-specific system can make the characteristics and behavior of provider 224, as well as provider signals 224a, transparent. Oracle 230 can then solve OSCRA 228 to provide OSC terms 226a using off-chain data at 232. Signals from receiver 222 can also be included at 222a. By observing a fairly accurate approximation of provider 224's actual effort, strong on-chain incentives can be created, making the cost of reputational damage for provider 224 high, for example, due to adverse behavior. In addition to reputation, an on-chain fund retention mechanism at the start of the contract can subsequently reward provider 224 if provider 224 exerts the most effort at 224b, or penalize (or reduce rewards) provider 224 if provider 224 exerts the least effort. Oracle 230 can also assist Provider 224 in delivering better services / goods or reducing the frequency and intensity of risks at 226b. Oracle 230 can also assess quality and help resolve any disputes (DAO governance can act as the final arbitrator if a complaint is filed by either the recipient or the provider). Recipient 222 can transact with Provider 224 at 226c using economic tokens applicable to DAO 202 or DApp 210. Provider 224 or Recipient 222 can purchase economic tokens or other tokens from or sell these tokens to DAO 202 at 224d and 222d respectively. Recipient 222 can pay its DAO 202 membership fee at 222b, for example, using tokens such as stablecoins or economic tokens, or a combination thereof.
[0223] For each DApp, the initial conditions of the existing mechanisms can be studied (e.g., tuples of existing probability measures when the ecosystem is not being used). This initial condition makes the receiver Initial loss occurs :
[0224]
[0225] Then, the OSC that minimizes the receiver's expected loss can be solved:
[0226] .
[0227] In this way, a value G is created for the receiver, which can represent the minimum loss resulting from the initial loss L0 and OSC. The difference between them. G can have a positive or zero value: zero when the contract is already Pareto optimal, and positive when the contract is initially in a Pareto suboptimal state.
[0228] The proportion of the value G created can be used Reward agreement. Additionally, it can be done through a ratio. A portion of the value created is reallocated to reward the most efficient provider. The value acquired by the protocol can be... The total compensation to the provider can be .
[0229] According to the embodiments described herein, a member recipient of a DApp may then become a provider of a product that it has received and subsequently modified or developed in another OSC entered into with other recipients, which may be located in the same DApp or another DApp. For example:
[0230] • The buyer of intermediate products can be the recipient, and can sell the final product, and can also become the supplier of goods in another optimal smart contract.
[0231] • The client of the outsourced service can be the recipient and can also become the service provider in another optimal smart contract.
[0232] • The buyer of real estate can be the recipient and can also become a debt provider in another DApp.
[0233] • The client of a risk prevention and remediation service provider can be a recipient and can also become a risk provider in another DApp.
[0234] Zero-knowledge proof identity authentication can be used to verify the human attributes and uniqueness of DAO members, for example, to defend against Sybil attacks. Unless there is a large-scale vote by DAO governance in the event of system abuse, it may be impossible to recover the identities of DAO members. Governance of the DAO can be structured to be fully decentralized using governance protocols, for example, by controlling the treasury of economic tokens, and by granting greater voting power to more senior DAO members. Therefore, trust can be a function of time.
[0235] In some implementations, DAO members earn one usage token (a non-transferable NFT) for each full year of use of at least one DApp. When a DAO member actively uses at least one DApp, they earn one governance token for each usage token they collect; this governance token is a non-transferable voting NFT. Therefore, voting power grows over time, rather than through the purchase of more tokens. This prevents the malicious seizure of control through the purchase of economic tokens, or the vote-buying seen in some other crypto-asset protocols.
[0236] Governance authority can be allocated at each DApp level by introducing dedicated governance tokens, for example, to handle feature requests. These tokens can also alleviate voting fatigue if DAO members are required to vote on DApps unrelated to them, for example, in DAO member voting.
[0237] Therefore, the fully decentralized governance of a DAO can be entirely controlled by real humans who can determine the strategic direction, scope, values (e.g., decentralization, privacy, fairness) of DApp development, as well as how externalities are handled (full privacy and ownership of DAO members' data, environmental impact, etc.). DApp development can heavily rely on artificial intelligence technologies (machine learning, deep learning) and has the potential to challenge the market for conventional or "Web2" AI applications. These centralized applications (CApps) can meet the need to efficiently create value for their end users, but may also be inherently detrimental to their privacy, which could be a critical externality. The DAO implementation described in this paper can leverage fully decentralized governance controlled entirely by real humans while maintaining unchanged values, thereby ensuring secure AI applications that deliver Pareto optimality for all value exchanges in a way that serves humanity.
[0238] In addition to utility tokens, board tokens, and governance tokens, which are all non-economic tokens, economic tokens can also be issued. We will now discuss various aspects of economic tokens and their minting process.
[0239] The supply side can include a minting supply of up to 21 million economic tokens. The majority of this minting supply can be based on usage, including 11 million economic tokens minted incrementally with adoption rates; and in terms of usage, the final economic token will be minted when the total usage time of the ecosystem exceeds 11 billion years. This is equivalent to issuing 11 billion usage tokens, assuming 1 billion DAO members use at least one DApp in the ecosystem for an average of 11 years. The second part of the minting process can be time-based: starting from the DAO's launch date, 10 million economic tokens can be minted in a geometric progression with a common ratio of 4 / 5 and a yearly cycle. Therefore, the minting function can be two-dimensional.
[0240] Figure 3 A graph 300 illustrating a casting function according to one or more embodiments of this document is shown. Figure 3 In the diagram, axis 310 corresponds to the total minted supply of economic tokens, axis 320 corresponds to usage, and axis 330 corresponds to time (measured in 5-year increments).
[0241] The supply side can also include minting and allocation. Whether issued based on usage or time, half of the allocation (10.5 million economic tokens) can be allocated to the DAO's community treasury, which can hold a significant proportion of these minted tokens. The other half of the tokens issued based on usage (5.5 million economic tokens) can be allocated to reward the community, for example, to incentivize DAO members or providers to develop markets capable of establishing optimal smart contracts, and also to reward DAO members for exercising DAO governance rights. The other half of the tokens issued based on time (5 million economic tokens) can be used to reward stakeholders such as contributors, intellectual property, long-term investors, and auditors.
[0242] In contrast to circulation, a portion of the minted supply can be dynamically and directly sold (without vesting period) to DAO members, providers, and investors in exchange for stablecoins, thereby financing the DAO and expanding DApp development for all different (industry) use cases.
[0243] Of the compensation awarded to contributors, intellectual property owners, and long-term investors, half can be vested linearly over time, for example, up to 10 years; and the other half can be released linearly between the current usage duration and a target of multiplying the usage duration by 10. This incentivizes contributors, researchers, and long-term investors to act in line with the DAO's goal of scaling globally.
[0244] Usage can be a major driver of demand, as the token enables interaction with all DApps. This mechanism offers a unique value proposition: a marginal capture of the total value created combined with potentially very high growth rates; rising demand driven by DAO usage, coupled with a slow increase in supply and the absence of pre-minted tokens, could offer significant upside potential for early token adopters, especially DAO members, providers, and investors, and is highly attractive and secure for token holders.
[0245] After decades of potentially strong growth in the value of economic tokens (in other words, demand for economic tokens will exceed supply), long-term supply-demand equilibrium should be achieved through the following elements:
[0246] • A DAO's community treasury can use a portion of its economic token reserves for the DAO's liquidity pools. In some implementations, the minting process can also directly provide economic tokens to the liquidity pools.
[0247] • DAO liquidity pools allow the buying and selling of economic tokens using stablecoins, thus enabling staking.
[0248] • Bonding curves can be used, which can be a type of Automated Market Maker (AMM). Such bonding curves can include a mathematical function of price, which is a function of the quantity of economic tokens in circulation and the quantity of economic tokens and stablecoins available in the liquidity pool.
[0249] Therefore, a positive price difference can always exist between the selling union curve and the buying union curve, thus avoiding arbitrage against DAOs.
[0250] Therefore, the union curve can provide stability between supply and demand.
[0251] Figure 4 An operating environment 400 according to one or more embodiments of the present invention is illustrated. For example... Figure 4 As illustrated, the operating environment 400 may include participants including user (receiver or provider) device 410, network 420, node 430 having at least one computing resource (such as processor 432) and electronic storage device 434, and remote device 440.
[0252] User (receiver or provider) device 410 may include any of a variety of devices that a user (receiver or provider) can use to interact with node 430 via network 420, including, for example, servers, desktop computers, laptop computers, handheld computers, tablet computing platforms, netbooks, smartphones, game consoles and / or other computing platforms.
[0253] Network 420 may include any of a variety of devices configured to enable the device to communicate with other devices, such as via wired and / or wireless connections, for example, via the Internet and / or other networks using cellular hardware such as TCP / IP or supporting wired or wireless communication (e.g., cellular communication, 2G, 3G, 4G, 4G LTE, 5G, or Wi-Fi). For example, network 420 may include, for example, a receiver, transmitter, transceiver, modem, network interface card, and / or antenna.
[0254] Node 430 may include any of a variety of devices that can be configured to perform the implementations and methods disclosed herein and interact with user (receiver or provider) device 410 via network 420, including, for example, servers, desktop computers, laptop computers, handheld computers, tablet computing platforms, netbooks, smartphones, game consoles and / or other computing platforms.
[0255] Node 430 may include processor 432. Processor 432 may include one or more processors, for example, configured to execute machine-readable instructions for implementing all or part of the embodiments described herein. Processor 432 may be configured to access electronic storage device 434 to retrieve electronic data from electronic storage device 434 and / or write electronic data to electronic storage device 434.
[0256] Node 430 may include electronic storage device 434. Electronic storage device 434 may be configured to electronically store (e.g., in a hosted manner) data corresponding to one or more databases or other data storage formats used in the embodiments herein. Processor 432 may be able to access electronic storage device 434.
[0257] Remote device 440 may include any of a variety of devices that a user (receiver or provider) can use to interact with node 430 via network 420, including, for example, servers, desktop computers, laptop computers, handheld computers, tablet computing platforms, netbooks, smartphones, game consoles and / or other computing platforms.
[0258] One or more additional nodes, such as node 450, can communicate electronically with one or more of the user (receiver or provider) device 410, remote device 440, and node 430 via network 420. Node 450 may have similar aspects and capabilities to, for example, node 430; that is, node 450 may also store part or all of the blockchain in a database on it to form a decentralized network that includes at least nodes 430 and 450.
[0259] In one example, if a user (receiver or provider) of user (receiver or provider) device 410 and a user (receiver or provider) of remote device 440 seek to establish an OSC between them, they can initially access the network through node 430. It should be understood that either or both of user (receiver or provider) device 410 or remote device 440 can also be nodes in a decentralized network, or they can interact with nodes (as described in this example).
[0260] Therefore, processor 432 can receive a set of mechanisms. Each mechanism can be defined at least by a provider action and a feasible contract. Processor 432 can define a receiver loss function based on the set of mechanisms. Processor 432 can determine the mechanism in the set that minimizes the receiver loss function. Processor 432 can generate a smart contract based on a mechanism. Processor 432 can broadcast the smart contract to multiple nodes via a network interface on network 420, and deploy the smart contract to a blockchain maintained on multiple nodes (e.g., including nodes 430 and 450) connected via computer network 420. Processor 432 can receive transaction addresses corresponding to the smart contract from the blockchain. Processor 432 can store the transaction addresses in a database of electronic storage device 434 and / or receiver device 410.
[0261] In some implementations, the processor 432 or the processor of the receiving device may also be configured to verify the smart contract by sending a data request to the blockchain in response to a transaction address.
[0262] Processor 432 can also be configured to receive oracle data. Each mechanism can also be defined by oracle data. For example, oracle data can be generated by sending a hint to a trained machine learning model and receiving the output of the trained machine learning model as oracle data. Receiving the output of the trained machine learning model can include performing data operations on the raw output data of the trained machine learning model, such as transformation, filtering, sorting, another data operation, or a combination thereof, to produce oracle data. Oracle data can be configured by hints received from a provider node or a receiver node.
[0263] In some implementations, transactions corresponding to OSCs may require, for example, a processor to transfer tokens to the blockchain. These tokens may be non-fungible tokens or non-destroyable tokens.
[0264] Each mechanism can also be defined by provider type, provider signal, and provider reward.
[0265] Each viable contract can also be defined by provider actions through provider signals and mechanisms. Each viable contract can also be defined by provider rewards. These provider rewards can be selected by the user (e.g., the receiver or the provider) from a set of possible provider rewards. A set of possible provider rewards can be selected based on provider signals.
[0266] Each viable contract may also be subject to one or more of the following constraints: incentive compatibility constraint, individual rationality constraint, or solvency constraint.
[0267] Processor 432 can also be configured to determine provider verification data and deploy it to the blockchain. Provider verification data may include, for example, provider reputation or certificate ratings. Processor 432 can retrieve available provider verification data from the blockchain, and each mechanism can also be defined by the provider verification data.
[0268] Although the functions described herein are performed by the hardware of node 430 (e.g., processor 432 and electronic storage device 434), it should be understood that the above functions can be implemented using any hardware or any combination of hardware from various devices (e.g., node 430, node 450, user (receiver or provider) device 410 or remote device 440).
[0269] Figure 5 This is a schematic diagram of example components of a device 500 according to one or more embodiments of this document. Device 500 may correspond to... Figure 4 One or more devices, networks, resources, or services. In some implementations, Figure 4 One or more devices, networks, resources, or services in device 500 may include one or more devices and / or components of device 500, for example, configured according to a client / server architecture, peer-to-peer architecture, and / or other architectures that may include multiple hardware, software, and / or firmware components that work together to provide the functionality attributable herein to device 500. In some implementations, device 500 may include a distributed computing architecture (e.g., one or more individual computing platforms that work together to perform computing tasks). For example, device 500 may be implemented by a cloud that works together as a computing platform for device 500. As a non-limiting example, a given device 500 may include one or more of servers, desktop computers, laptop computers, handheld computers, tablet computing platforms, netbooks, smartphones, game consoles, and / or other computing platforms.
[0270] As used herein, the term "component" is intended to be interpreted broadly as hardware, firmware, or a combination of hardware and software. Clearly, the systems and / or methods described herein can be implemented using various forms of hardware, firmware, and / or combinations of hardware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods does not limit the implementation. Therefore, since the operation and behavior of systems and / or methods are described herein without reference to specific software code, it should be understood that the systems and / or methods can be implemented using software and hardware based on the description herein.
[0271] Device 500 may include bus 510, processor 520, memory 530, storage unit 540, input unit 550, output unit 560 and communication unit 570.
[0272] Bus 510 includes components that enable wired and / or wireless communication between parts of device 500. Bus 510 enables various parts of a computer system to communicate with each other, thereby allowing data to be transferred from one part to another.
[0273] Processor 520 may include a central processing unit, a graphics processing unit, a microprocessor, a controller, a microcontroller, a digital signal processor, a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), and / or another type of processing unit. Processor 520 may be implemented in hardware, firmware, or a combination of hardware and software. In some embodiments, processor 520 may include one or more processors that can be programmed to perform functions. These processors may or may not be all integrated in the same physical device, and in some embodiments may be distributed across multiple devices.
[0274] Processor 520 may be configured to execute one or more modules and / or other modules disclosed herein through software, hardware, firmware, a combination of software, hardware and / or firmware, and / or other mechanisms for configuring the processing capabilities of processor 520. As used herein, the term "module" may refer to any part or collection of parts that performs the functions belonging to that module. This may include one or more physical processors, processor-readable instructions, circuitry, hardware, storage media, or any other part during the execution of processor-readable instructions. Various modules or portions thereof may be implemented in any of a variety of ways, including process-based techniques, component-based techniques, and / or object-oriented techniques. For example, program instructions can be implemented using system libraries, language libraries, Model-View-Controller (MVC) principles, Application Programming Interfaces (APIs), Large Language Models (LLMs), system-specific programming languages and principles, cross-platform programming languages and principles, pre-compiled programming languages, markup programming languages, stylesheet languages, "bytecode" programming languages, object-oriented programming principles or languages, other programming principles or languages, C, C++, C#, Java, JavaScript, Python, PHP, HTML, CSS, TypeScript, R, Elm, Unity, VB.Net, Visual Basic, Swift, Objective-C, Perl, Ruby, Go, SQL, Haskell, Scala, Arduino, assembly language, Microsoft Foundation Classes (MFC), Streaming SIMD Extensions (SSE), or other technologies or methods as needed.
[0275] It should be understood that although some modules disclosed herein may be exemplified, for example, as being implemented within a single processing unit, in embodiments where processor 520 includes multiple processing units, one or more of the modules disclosed herein may be implemented remotely relative to other modules. The descriptions of the functionality provided by different modules herein are for illustrative purposes only and are not intended to be limiting, as any module described herein may provide more or fewer functions than described. For example, one or more modules disclosed herein may be removed, and some or all of their functionality may be provided by other modules disclosed herein. Furthermore, processor 520 may be configured to execute one or more additional modules that may perform some or all of the functionality of one of the modules categorized as modules disclosed herein.
[0276] The memory 530 may include random access memory, read-only memory and / or another type of memory (e.g., flash memory, magnetic memory and / or optical memory).
[0277] Electronic storage component 540 may store information and / or software related to the operation of device 500. For example, electronic storage component 540 may include a solid-state drive, hard disk drive, disk drive, optical disk drive, compact disk, digital multifunction disk, and / or another type of non-transitory computer-readable medium. Embodiments of electronic storage component 540 may include one or more of optically readable storage media (e.g., optical discs, etc.), magnetically readable storage media (e.g., magnetic tape, magnetic hard disk drive, floppy disk drive, etc.), charge-based storage media (e.g., EEPROM, RAM, etc.), solid-state storage media (e.g., flash memory drives, etc.), and / or other electronically readable storage media. Embodiments of electronic storage component 540 may include one or both of the following: a system storage device integrated into device 500 (i.e., substantially non-removable); and / or a removable storage device removably connectable to device 500, for example, via a port (e.g., serial port, USB port, IEEE 1394 port, THUNDERBOLT™ port, etc.) or a drive (e.g., disk drive, flash memory drive, or solid-state drive, etc.). Electronic storage component 540 may also, or alternatively, include one or more virtual storage resources (e.g., cloud storage, virtual private networks, and / or other virtual storage resources). The electronic storage device may store software algorithms, information determined by one or more processors, information received from one or more computing platforms, information received from one or more remote platforms, databases (e.g., Structured Query Language (SQL) databases (e.g., MySQL®, MARIADB®, MONGODB®), NoSQL databases, etc.), data files, compiled data, analytical data, charts, tables, videos, images, presentations, and 3D content in appropriate formats, and / or other information that enables the computing platform to operate as described herein.
[0278] Electronic storage component 540 can host distributed ledgers, such as blockchains. Blockchains can be used to maintain a continuously growing list of records called blocks. Blockchains can be used to maintain records of transactions or events between parties in a way that is difficult to forge. Each block in a blockchain can include several records as well as hashes of previous blocks in the blockchain. If a record in a previous block is changed, the hash in any subsequent block can be corrupted. Therefore, the system can easily detect whether the fidelity of the data is maintained. Similarly, to forge a given record, a malicious actor would need to forge that record as well as all subsequent records, ensuring that the final hash remains the same. This is extremely difficult in practice. Furthermore, blockchains can be distributed among multiple entities. Any change to the blockchain can be verified by comparing it to a large number of individual records.
[0279] Input component 550 enables device 500 to receive input, such as user (receiver or provider) input and / or sensed input. For example, input component 550 may include a touchscreen, keyboard, keypad, mouse, button, microphone, switch, sensor (internal and / or external), GPS component, accelerometer, gyroscope and / or actuator.
[0280] The output component 560 enables the device 500 to provide output, such as via a display, speaker, and / or one or more light-emitting diodes.
[0281] Communication component 570 enables device 500 to communicate with other devices, such as via wired and / or wireless connections, for example, via the Internet and / or other networks using cellular hardware that supports wired or wireless communication (e.g., cellular communication, 2G, 3G, 4G, 4G LTE, 5G, Wi-Fi, Near Field Communication (NFC), BLUETOOTH®). For example, communication component 570 may include a receiver, transmitter, transceiver, modem, network interface card, and / or antenna.
[0282] As used herein, the “Internet” can include an interconnected network of systems and a set of protocols for end-to-end data transfer between these systems. A model describing the Internet can be the Transmission Control Protocol and the Internet Protocol (TCP / IP), also known as the Internet Protocol Suite. TCP / IP provides a four-layer abstraction model: the application layer, the transport layer, the Internet layer, and the data link layer. The data link layer can include hosts that can be accessed without going through a router and can therefore be determined by the network configuration (e.g., hardware network implementation, local area network, virtual private network, or network tunnel). The data link layer can be used to move data packets between Internet layer interfaces of different hosts on the same link. The data link layer can interact with hardware used for end-to-end data transfer. The Internet layer can include datagram exchange across network boundaries (e.g., from a source network to a destination network), which may be referred to as routing and is performed through host addressing and identification in the Internet Protocol (IP) addressing system (e.g., IPv4, IPv6). A datagram can include a self-contained, independent basic data unit containing a header (e.g., including source address, destination address, and type) and a payload (e.g., the data to be transmitted) for transmission across packet-switching networks. The transport layer can use User Datagram Protocol (UDP) to provide a basic data channel (e.g., via a network port) for applications to exchange data, establishing end-to-end, host-to-host connections that are independent of any underlying network or the structure of the user data. The application layer can include various user protocols and supporting protocols used by applications to create and exchange data, use services, or provide services over network connections established by lower layers. These protocols include, for example, routing protocols, Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), Simple Mail Transfer Protocol (SMTP), and Dynamic Host Configuration Protocol (DHCP). This data creation and exchange in the application layer can, for example, use a client-server model or a peer-to-peer network model. Data from the application layer can be encapsulated in UDP datagrams or TCP streams to interact with the transport layer, which then transmits the data via lower layers.
[0283] The communication component 570 can also implement an Internet of Things (“IoT”) configuration, which may include a network of physical objects (devices, vehicles, buildings, and other items) embedded with electronics, software, sensors, and network connectivity, enabling these objects to collect and exchange data via the Internet. Each IoT product / device can be an endpoint device with its own Internet address (e.g., IPv4, IPv6 address). IoT allows for remote sensing and control of objects across existing network infrastructure (e.g., the Internet), creating opportunities to integrate the physical world more directly into computer-based systems.
[0284] Device 500 can perform one or more processes described herein. For example, a non-transitory computer-readable medium (e.g., memory 530 and / or storage component 540) can store a set of instructions (e.g., one or more instructions, code, software code, and / or program code) for execution by processor 520. Processor 520 can execute a set of instructions to perform one or more processes described herein. In some embodiments, execution of a set of instructions by one or more processors 520 causes one or more processors 520 and / or device 500 to perform one or more processes described herein. In some embodiments, hard-wired circuitry may be used instead of instructions or in combination with instructions to perform one or more processes described herein. Therefore, the embodiments described herein are not limited to any particular combination of hardware circuitry and software.
[0285] Figure 5 The number and arrangement of components shown are provided as an example only. Device 500 may include more than Figure 5 The device 500 may have more components, fewer components, different components, or components arranged differently. In addition or alternatively, a group of components (e.g., one or more components) of the device 500 may perform one or more functions described as being performed by another group of components of the device 500.
[0286] remove Figure 5 Beyond the example configurations described herein, various steps, functions, and / or operations of device 500, as well as the methods disclosed herein, can be executed by one or more of, for example, electronic circuits, logic gates, multiplexers, programmable logic devices, ASICs, analog or digital controllers / switches, microcontrollers, or computing systems. Program instructions implementing methods such as those described herein can be transmitted or stored on a carrier medium. The carrier medium may include storage media such as read-only memory, random access memory, magnetic disks or optical disks, non-volatile memory, solid-state memory, magnetic tape, etc. The carrier medium may also include transmission media such as wires, cables, or wireless transmission links.
[0287] Figure 6 This is a schematic diagram of example components of device 670 according to one or more embodiments of this document. Device 670 may correspond to communication component 570 or... Figure 4 or Figure 5 Another device or component illustrated. In some embodiments, communication component 570 or... Figure 4 or Figure 5 Another illustrated device or component may include one or more devices and / or one or more components of device 670. For example... Figure 6As shown, device 670 may include one or more input components 672 (collectively referred to herein as input component 672 or simply input component 672), switching component 674, one or more output components 676 (collectively referred to herein as output component 676 or simply output component 676), and controller 678.
[0288] Input component 672 may be one or more connection points for one or more input physical links 671 (collectively referred to herein as input physical link 671 or simply input physical link 671), and includes one or more entry points for incoming traffic (such as packets). Input component 672 may process incoming traffic, such as by performing data link layer encapsulation or decapsulation. In some embodiments, input component 672 may send and / or receive packets. In some embodiments, input component 672 may include an input line card that includes one or more packet processing components (e.g., in the form of an integrated circuit), such as one or more interface cards (IFCs), packet forwarding components, line card controller components, input ports, processors, memory, and / or input queues. In some embodiments, device 670 may include one or more input components 672.
[0289] Switching unit 674 interconnects input unit 672 with output unit 676. In some embodiments, switching unit 674 may be implemented via one or more crossbar switches, via a bus, and / or utilizing shared memory. Shared memory may be used as a temporary buffer to store packets from input unit 672 before final scheduling and delivery of packets to output unit 676. In some embodiments, switching unit 674 enables input unit 672, output unit 676, and / or controller 678 to communicate with each other.
[0290] Output component 676 can store packets and schedule packets for transmission on output physical link 679 (collectively referred to herein as output physical link 679 or solely as output physical link 679). Output component 676 can support data link layer encapsulation or decapsulation and / or various higher-layer protocols. In some embodiments, output component 676 can transmit and / or receive packets. In some embodiments, output component 676 may include an output line card that includes one or more packet processing components (e.g., in the form of an integrated circuit), such as one or more IFCs, packet forwarding components, line card controller components, output ports, processors, memory, and / or output queues. In some embodiments, device 670 may include one or more output components 676.
[0291] In some implementations, the input component 672 and the output component 676 may be implemented by the same set of components (e.g., the input / output component may be a combination of the input component 672 and the output component 676).
[0292] Controller 678 includes a processor, for example, in the form of a CPU, GPU, APU, microprocessor, microcontroller, DSP, FPGA, ASIC, and / or another type of processor. The processor is implemented in hardware, firmware, or a combination of hardware and software. In some embodiments, controller 678 may include one or more processors that can be programmed to perform functions.
[0293] In some implementations, controller 678 may include RAM, ROM and / or another type of dynamic or static storage device (e.g., flash memory, magnetic storage, optical storage, etc.) to store information and / or instructions for use by controller 678.
[0294] In some implementations, controller 678 can communicate with other devices, networks, and / or systems connected to device 670 to exchange information about the network topology. Controller 678 can create routing tables based on the network topology information, create forwarding tables based on the routing tables, and forward the forwarding tables to input unit 672 and / or output unit 676. Input unit 672 and / or output unit 676 can use the forwarding tables to perform route looks for incoming and / or outgoing packets.
[0295] Controller 678 can perform one or more of the processes described herein. Controller 678 can perform these processes in response to executing software instructions stored on a non-transitory computer-readable medium. Computer-readable medium is defined herein as a non-transitory storage device (e.g., "non-transitory" is a limitation on the medium itself (i.e., a tangible medium, not a signal), rather than a limitation on the persistence of data storage (e.g., RAM and ROM)). Storage devices include storage space within a single physical storage device or storage space distributed across multiple physical storage devices.
[0296] Software instructions can be read from another computer-readable medium or another device via a communication interface into the memory and / or storage unit associated with controller 678. When executed, the software instructions stored in the memory and / or storage unit associated with controller 678 can cause controller 678 to perform one or more processes described herein. Alternatively or additionally, hardwired circuitry can be used in place of or in combination with software instructions to perform one or more processes described herein. Therefore, the embodiments described herein are not limited to any particular combination of hardware circuitry and software.
[0297] Figure 6The number and arrangement of components shown are provided as an example only. In practice, device 670 may include more than Figure 6 The device 670 may have more components, fewer components, different components, or components arranged differently. In addition or alternatively, a group of components (e.g., one or more components) of the device 670 may perform one or more functions described as being performed by another group of components of the device 670.
[0298] The following figures illustrate example methods and their operation. In some embodiments, the methods illustrated herein may include more operations, fewer operations, operations arranged differently, or different operations than those depicted in the following figures. Furthermore, or alternatively, two or more operations depicted in the following figures may be performed at least partially in parallel.
[0299] In embodiments of the methods illustrated in the following figures, various operations may be performed by one or more hardware processors configured with machine-readable instructions (e.g., instructions electronically stored on an electronic storage medium), which may include modules according to one or more embodiments. Such hardware processors may include one or more processing devices (e.g., one or more digital processors, analog processors, digital circuits designed to process information, analog circuits designed to process information, state machines, and / or other mechanisms for electronically processing information). The one or more processing devices may include one or more devices configured by hardware, firmware, and / or software, which may be specifically designed to perform one or more operations of the methods illustrated herein.
[0300] Figure 7 This is a flowchart illustrating an example method 700 according to one or more embodiments of this document. In some embodiments, Figure 7 One or more of the illustrated operations can be performed by Figures 2 to 6 One or more of the depicted devices or components perform operations collaboratively, alternatively, or in combination. In some embodiments, one or more operations may be performed by another device, system, or group of devices or systems that is separate from or includes these devices or systems. Additionally or alternatively, other devices, components, or systems may be used to perform these operations.
[0301] Operation 702 may include: receiving a set of mechanisms at the processor of one of a plurality of nodes, wherein each mechanism is defined at least by a provider action and a feasible contract; and the operation may be performed individually or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0302] Operation 706 may include: using a processor to determine a mechanism from a set of mechanisms that minimizes the receiver's loss function; and this operation may be performed alone or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0303] Operation 708 may include: using a processor to generate smart contracts based on a mechanism; and this operation may be executed alone or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0304] Operation 710 may include: using a processor to broadcast a smart contract to multiple nodes via a network interface that communicates electronically with the processor and a computer network, thereby deploying the smart contract to a blockchain maintained across multiple nodes connected via a computer network; and this operation may be performed individually or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0305] Operation 712 may include: receiving a transaction address corresponding to a smart contract from the blockchain at the processor; and this operation may be performed alone or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0306] Operation 714 may include: using a processor to store transaction addresses on the node's electronic storage device and in a database that communicates electronically with the processor; and this operation may be performed independently or in conjunction with... Figure 7 One or more other operations described are performed in combination.
[0307] Implementations of this disclosure may include collaborative use cases, such as for collaborative work (“optimal work”). In this example:
[0308] • Providers: Colleagues, employees, freelancers
[0309] • Product: Work
[0310] • Recipient: Project Owner
[0311] Problem: Information asymmetry and inconsistency between employers and employees is a significant challenge in the workplace. It can lead to misunderstandings, dissatisfaction, and inefficiency. This problem manifests as inconsistent expectations, communication gaps, lack of transparency, skills and training mismatches, and ineffective feedback mechanisms. From an economic perspective, it exemplifies the principal-agent problem in the workplace and can lead to adverse selection and moral hazard. This results in inefficiency and leads to high opportunity costs and insufficient investment in human capital.
[0312] Instructions for use with OSCRA:
[0313] • Type indicators / variables: role, education level, skills, experience
[0314] • Action indicators / variables: performance, trust / reliability
[0315] • Signal indicators / variables: Achieved milestones, team well-being
[0316] Compensation metrics / variables: hourly rate, bonuses / penalties
[0317] In its initial state, the Optimal Work application lacks an OSC (Optimal Smart Contract), and there is no information on the provider's performance and credibility in the market. Any new provider joining the ecosystem can have its initial performance rating and credibility value initialized to 0.5 (a standard continuous uniform distribution between 0 and 1). The contributor's initial hourly rate corresponds to the 50th percentile (median) of the salary distribution obtained from statistical market data (e.g., retrieved from salary and market data via the internet), corresponding to the factors explaining the salary level for that role (role, location, etc.).
[0318] Salaries are paid to the work provider (contributor) using a given asset (such as cash) based on the agreed hourly rate and number of hours worked. Contributors record their actual hours worked in a reporting tool, which is then approved by the project manager. Over time, the contributor's rating changes based on a list of metrics, such as performance indicators and credibility metrics.
[0319] Implementations of this disclosure may include asset tracking and trading (“optimal assets”). In this example:
[0320] • Provider: The issuer of these assets (stocks / tokens / commodities) who possesses privileged information regarding the risks (operational risks, credit risks, etc.) and potential of the underlying assets.
[0321] • Product: Fair and risk-adjusted valuation of assets
[0322] • Recipient: The investor of the asset
[0323] Instructions for use with OSCRA:
[0324] • Type indicators / variables: Asset risk level, asset class (innovation / potential, sustainability, ESG, etc.), team potential, etc.
[0325] • Action indicators / variables: performance, ESG score
[0326] • Signal indicators / variables: Milestones achieved by the project
[0327] • Return indicator / variable: Cash payment for asset purchase
[0328] Implementations of this disclosure may include investment management (“Optimal Investment”). Optimal Investment is a solution for retail investors that enables them to invest in portfolios under a self-custodial model and achieve Pareto optimal asset allocation (a smart contract-based asset management tool based on their individualized risk profile—assets can be Bitcoin, alternative currencies, stablecoins, and commodities such as gold, or other RWA tokenized stocks). It protects retail users from cryptocurrency scams through a self-custodial asset management solution aligned with their risk appetite. It is an optimal DApp with the following key roles and characteristics:
[0329] • Provider: The asset manager, who has more information about market forces (holders and short sellers of these assets) and may collude with the issuers of these assets (e.g., colluding on the market order book).
[0330] • Products: Assets in the investment portfolio
[0331] • Recipients: Investors of these assets
[0332] Currently, issuer credit risk and pump-and-dump schemes implemented by market forces can create unfairness. Both types of participants lack transparency and incentives for good behavior. This solution can provide recipients with transparency, verifiability, and ultimately fairness, as well as actual ownership of their asset strategies.
[0333] Problem: The current portfolio provider or issuer possesses privileged information regarding a variety of asset options (such as stocks, forex, contracts for difference (CDF), ETFs, commodities, bonds, mutual funds, cryptocurrencies, tokenized stocks, etc.) and specific market forces (e.g., holders and short sellers of these assets), resulting in contractual imbalances. Credit risk from the issuer or pump-and-dump schemes implemented by market forces may contribute to this unfairness.
[0334] In this example, the following defects mentioned above are resolved:
[0335] • Calculate the Pareto optimal asset allocation based on the members' risk preferences and the actual risk of each investment asset in the market.
[0336] Pareto optimal investing also incorporates a weighted fixed-amount investment (DCA) approach based on current asset risk and members' risk aversion to smooth out market volatility-related risks: in each period (e.g., weekly), the allocation is automatically rebalanced on-chain in a self-custodial manner to bring it closer to the Pareto optimal asset allocation.
[0337] Even assuming that members' risk aversion remains constant, Pareto optimal asset allocation will change based on the market risk of each asset (and especially its stage in the cycle). In this scenario, the onset of a recession will lower the prices of risky assets such as Bitcoin and Ethereum: if their long-term fundamental value remains unchanged, allocations to these assets should be increased.
[0338] • Over time, the current configuration converges to the Pareto optimal configuration.
[0339] The best investment is both the best DApp and an asset management tool.
[0340] Our smart contract-based asset management tools with Pareto optimality algorithms are designed to meet investors' individual risk profiles in a transparent, decentralized manner, unaffected by the hidden motives of issuers.
[0341] This on-chain asset management tool also employs a DCA approach based on current asset risk and investor risk aversion, which means that allocations are automatically adjusted to smooth out market volatility-related risks.
[0342] Investors can access optimal investment tools by paying a DAO membership fee.
[0343] This example provides:
[0344] • Customization: Customized risk profile with Pareto optimal asset allocation
[0345] • True ownership: A portfolio controlled by investors and securely existing on-chain (immutable) in a self-custodial manner.
[0346] • Risk transparency: Investors have access to the same risk information as the issuer.
[0347] • Native on-chain weighted DCA method: Automatic periodic investment to save costs and minimize risk.
[0348] • Competitive membership fees: Our business model is competitive with typical asset management fees, thereby lowering the barrier to entry for users and promoting large-scale adoption.
[0349] • Decentralization and governance: Investors can participate in the governance of the upper-level DAO (the best DAO) and express their opinions on how the services they use are formed.
[0350] Regarding the process operation mode for the recipient:
[0351] • Joining process: Connect on the blockchain (ICP) and verify identity on Decide AI to obtain a unique ID.
[0352] Register and specify the amount of assets to be included in management.
[0353] ·answer the questions:
[0354] Choose assets X, Y, Z from Bitcoin and a fiat stablecoin (such as USDT used for POC) to invest in (then gradually add crypto gold, other fiat stablecoins, and Ethereum for the first iteration of MVP).
[0355] ○ Select the initial amounts Q0_X, Q0_Y, and Q0_Z to be deposited into the smart contract (in a self-managed manner).
[0356] Is it possible that in the future, at any time, some of assets X, Y, or Z will need to be redeemed from the smart contract with a probability of over 95% (or other probability)? -> We will add a disclaimer assuming no liability. If so, the investment strategy should be strictly controlled and highly stable on that asset to ensure a greater than 95% certainty of redemption at any time.
[0357] ○Then provide some scenarios, such as: "If X, Y, and Z lose 50% on average in one of the four years, but the target average annual return is 30%, would you be able to accept this?" to assess the members' risk aversion level.
[0358] ○ We agree with the main model assumptions of our algorithm regarding Pareto optimal asset allocation (Bitcoin outperforms the US dollar in the long run).
[0359] • Calculation of risk level:
[0360] ○ Initially conducted on a centralized server,
[0361] In the long term, this will be done on smart contracts (e.g., on ICPs).
[0362] • Deposit one or more managed assets
[0363] • Fees are paid based on assets under management (generally slightly less than 1% of total assets under management).
[0364] • Deposited funds
[0365] Instructions for use with OSCRA:
[0366] • Type indicators / variables: Consistency between the asset manager's underlying assumptions and the investor's assumptions; consistency between the risk level of the asset management strategy and the investor's risk preference; asset classes in the portfolio (innovation / potential, sustainability, ESG, etc.), etc.
[0367] • Action indicators / variables: Performance indicators based on previous asset allocation strategy performance or backtesting, and asset ESG scores.
[0368] • Signal indicators / variables: Portfolio profit and loss
[0369] Compensation metrics / variables: asset management fees, bonuses for good or bad performance, and risk sharing.
[0370] Initially, the affiliated company of the optimal DAO is the sole provider (asset manager). After the MVP phase, competition will be opened, allowing different providers to conduct business on the platform.
[0371] • In the initial state, the optimal investment application does not have an OSC (Optimal Smart Contract), and there is no information on the provider's (asset manager's) performance metrics in the market. Any new provider joining the ecosystem is initialized with an initial performance score and credibility value of 0.5 (a standard continuous uniform distribution between 0 and 1).
[0372] Embodiments of this disclosure may include medical procedures, such as dental procedures (“optimal dentistry”). In this example:
[0373] • Provider: Medical practitioners
[0374] Products: Diagnosis and Treatment
[0375] • Recipient: Patient
[0376] Secondary medical opinions can be provided based on AI models: scanning members' X-ray images and using a predictive model to provide a diagnosis, thereby protecting patients from overtreatment and executing optimal smart contracts in the medical field. During the algorithm's preparation / training phase (performed periodically), all members' X-ray images are scanned, and the predictive model provides a score of the practitioner's success rate. During the inference phase (whenever a member wishes to generate an OSC), the following data is already available:
[0377] • Regarding the provider's product and utility functions: quality rating, price, location, availability, user experience, etc.
[0378] • Utility function for the receiver (determined at join time)
[0379] Instructions for use with OSCRA:
[0380] • Type metrics / variables: dentist location, user experience (UX), usability, wait time, etc.
[0381] • Action indicators / variables: Diagnostic accuracy score, treatment success rate, trust / reliability score
[0382] • Signal indicators / variables: Best estimate of AI model (second opinion in diagnosis), UX rating
[0383] • Compensation Indicator / Variable: Diagnostic Price
[0384] Implementations of this disclosure may include a used car market (“optimal car”). In this example:
[0385] • Provider: Used car dealer
[0386] Product: Used cars
[0387] • Recipient: Used car buyer
[0388] Implementations of this disclosure may include risk sharing (“optimal risk-sharing pool”). In this example:
[0389] • Provider: Risk Provider / Insured
[0390] • Product: Risk Transfer
[0391] • Recipient: Risk bearer / insurer (mutual aid organization)
[0392] Instructions for use with OSCRA:
[0393] • Type indicator / variable: Risk level
[0394] • Action indicators / variables: Efforts to reduce risk exposure (frequency, intensity, etc.)
[0395] • Signal indicators / variables: The actual occurrence of risk
[0396] • Compensation indicator / variable: Insurance claim payments (minus premiums paid before the commencement of insurance liability)
[0397] Implementations of this disclosure may include dynamic token splitting for collaborative projects. This implementation may combine aspects of collaborative implementations and asset tracking and trading implementations (“Optimal Dynamic Token Splitting” or “ODTS”).
[0398] ODTS can be analogous to a combination of optimal work and optimal assets, used to ensure a fair and dynamic equity / token split among the co-founders of a project. It fairly redistributes shares / tokens based on a risk-adjusted valuation of the added value contributed by the co-founders to the project, thereby protecting startup co-founders from free-riders.
[0399] Implementations of this disclosure may include insurance. Such implementations may combine aspects of risk-sharing implementations and medical procedure implementations (“optimal insurance”).
[0400] Optimal insurance can be similar to a combination of optimal dental care (or optimal car insurance, etc.) and optimal risk pooling, thereby significantly reducing premiums in these markets. This is because the information asymmetry and inconsistency between the risk bearer and the third party, which form the basis for risk prevention, claims management and repair, has been resolved by an optimal DApp within the corresponding optimal DApp.
[0401] One implementation method is machine learning, a form of artificial intelligence (AI) that enables computers to learn how to process data without being explicitly programmed. Machine learning focuses on developing computer programs that can learn, grow, and change themselves when exposed to new data. It studies and builds algorithms that can learn from data and make predictions based on that data. By building models from sample inputs, these algorithms can make data-driven predictions or decisions, thus breaking free from the limitations of strictly static program instructions.
[0402] Machine learning can refer to various AI software algorithms that can be used to perform supervised learning, unsupervised learning, reinforcement learning, deep learning, or any combination thereof. A variety of different machine learning algorithms can be employed in the implementation. Examples of machine learning algorithms may include, in particular, artificial neural network algorithms, Gaussian process regression algorithms, fuzzy logic-based algorithms, or decision tree algorithms.
[0403] In some implementations, more than one machine learning algorithm may be employed. For example, one machine learning algorithm may be used for automatic classification, while different types of machine learning algorithms may be used for adaptive real-time process control. In some implementations, a hybrid machine learning algorithm may be employed, incorporating features and properties obtained from two, three, four, five, or more different types of machine learning algorithms.
[0404] Supervised learning algorithms can use labeled training data to infer the relationship between one or more identifiable aspects of a given entity and the classification of the entity according to a specified set of criteria, or to infer the relationship between input process control parameters and the desired outcome. Training data can include pairs of training examples. For example, each training data example can include aspects identified for a given entity and the final classification of the given entity. Alternatively, each training data example can include process control parameters used in the process and known results of the process.
[0405] Unsupervised learning algorithms can be used to make inferences based on training data that includes entity data not paired with labeled entity classification data, or input process control parameter data not paired with labeled process results. An example of an unsupervised learning algorithm is cluster analysis, which can be used for exploratory data analysis to discover hidden patterns or groupings in process data.
[0406] Semi-supervised learning algorithms can be trained using both labeled and unlabeled object classification data or process data. They typically utilize a small amount of labeled data and a large amount of unlabeled data.
[0407] Reinforcement learning algorithms can be used, for example, to optimize processes (e.g., process steps or actions) to maximize a process reward function or minimize a process loss function. In a machine learning context, reinforcement learning algorithms can be formulated as Markov decision processes. A reward function or loss function (also called a cost function or error function) maps the values of one or more process variables and / or outcomes to real numbers representing the reward or cost associated with a given process outcome or event. Examples of process parameters and outcomes include, in particular, process throughput, process yield, production quality, or production cost. In some cases, the definition of the reward function or loss function to be maximized or minimized can depend on the choice of the machine learning algorithm used to run the process control method, and vice versa. For example, if the objective is to maximize the total reward / value function, a reinforcement learning algorithm can be chosen. If the objective is to minimize the mean squared error loss function, a decision tree regression algorithm or a linear regression algorithm can be chosen. Typically, machine learning algorithms used to run process control methods seek to optimize the reward function or minimize the loss function by: identifying the current state of the process; comparing the current state with a reference state, which can be a target intermediate state or the final state; and adjusting one or more process control parameters to minimize the difference between the two states. This adjustment can reference past learning results provided by the training dataset. Reinforcement learning algorithms differ from supervised learning algorithms in that they do not provide correct training data input / output pairs, nor do they explicitly correct Pareto suboptimal actions. These algorithms are often implemented with a focus on real-time performance by striking a balance between exploring possible outcomes based on updated input data and utilizing past training results.
[0408] Deep learning, also known as deep structured learning, hierarchical learning, or deep machine learning, is based on a set of algorithms that attempt to model high-level abstractions in data. Deep learning algorithms can be inspired by the structure and function of the human brain and belong to a broader family of machine learning methods based on learning data representations. Deep learning is based on neural network techniques and can involve probabilistic graphical models with many layers of neurons, often referred to as deep architectures. Deep learning techniques can process information such as, in particular, images, text, or sound information in a hierarchical manner. Observations (e.g., features to be extracted for reference) can be represented in various ways, including, for example, intensity value vectors, edge sets, shape regions, or other abstractions. Some representations can simplify learning tasks (e.g., face recognition or facial expression recognition). Deep learning can provide efficient algorithms for unsupervised or semi-supervised feature learning and hierarchical feature extraction. Implementations using deep learning can also leverage the advantages of deep learning concepts in solving representation inversion problems that are typically difficult to handle.
[0409] Deep learning modules can be configured as neural networks. A deep learning module can also be a deep neural network, which has a set of weights that model the world based on training using training data. A neural network can be understood as implementing a computational method—based on a relatively large number of neural units—to roughly simulate how the human brain solves problems using large groups of biological neurons connected by axons. Each neural unit can connect to one or more other neural units, and these connections can enhance or inhibit the activation state of the connected neural units. These systems can learn and train themselves, rather than being explicitly programmed. Neural network systems are particularly adept at expressing solutions or feature detection in areas where traditional computer programs struggle.
[0410] An example of a deep learning algorithm is an artificial neural network (ANN). Large ANNs, which include many layers, can be used, for example, to map entity data to entity classification decisions or to map input process control parameters to desired process outcomes. ANNs will be discussed in more detail below.
[0411] Neural networks typically consist of multiple layers, and signal paths can traverse these layers from front to back. The goal of a neural network can be to solve a problem in a manner similar to the human brain, although some neural networks may be more abstract. In a simple example of a neural network, there can be two layers (i.e., two sets) of neurons: an input layer that receives input signals and an output layer that sends output signals. When the input layer receives input, it can pass a modified version of the input to the next layer. In deep networks, there can be many layers between the input and output layers, allowing the algorithm to use multiple processing layers, which can include multiple linear and nonlinear transformations. Modern neural networks typically utilize thousands to millions of neurons and millions of connections to operate. Neural networks can have a variety of suitable architectures and / or configurations known in the art.
[0412] Based on probabilistic norms and network architectures, deep neural networks have many variations, including deep belief networks (DBNs), restricted Boltzmann machines (RBMs), random forests, and autoencoders. The implementation of neural networks can vary depending on the size of the input data, the number of features to be analyzed, and the nature of the problem. In addition to the neural networks disclosed herein, other layers can also be included in deep learning modules.
[0413] Another type of deep neural network can be a convolutional neural network (CNN), which can be used to analyze entities or processes. CNNs typically consist of different types of layers: convolutional layers, pooling layers, upsampling layers, and fully connected layers. In some cases, activation functions, such as the rectified linear unit (ReLU) function, can be used in some of the layers. In a CNN architecture, there can be one or more layers for each type of operation performed. A CNN architecture can include any total number of layers, and any number of layers for performing different types of operations. The simplest CNN architecture begins with an input layer, followed by a sequence of convolutional and pooling layers (e.g., layers that are separately constructed to reduce the dimension of the feature maps generated by one or more convolutional layers while preserving the most important features, such as max pooling layers), and ends with fully connected layers (e.g., a layer where each node in a node is connected to every node in the previous layer). Each convolutional layer can include multiple parameters for performing the convolution operation. Each convolutional layer can also include one or more filters, which in turn can include one or more weight factors or other adjustable parameters. In some cases, these parameters can include biases (e.g., parameters that allow the activation function to shift). In some cases, a ReLU activation function layer can follow a convolutional layer. Other activation functions can also be used, such as the saturated hyperbolic tangent, identity function, binary step function, logistic function, arctangent function, softsign function, parameterized modified linear unit function, exponential linear unit function, softPlus function, curved identity function, softExponential function, sine function, Sinc function, Gaussian function, or sigmoid function. Convolutional layers, pooling layers, and ReLU layers can be used as learnable feature extractors, while fully connected layers can be used as machine learning classifiers. Like other artificial neural networks, the convolutional and fully connected layers of a CNN architecture can include various computational parameters, such as weights, biases, and thresholds, which can be trained during the training phase.
[0414] Another type of deep neural network can be the Visual Geometric Group (VGG) network. For example, a VGG network can be created by increasing the number of convolutional layers while keeping other parameters of the architecture fixed. Depth can be increased by adding convolutional layers using essentially small convolutional filters in all layers. A VGG network can also consist of convolutional layers followed by fully connected layers.
[0415] Another type of deep neural network can be a deep residual network. Similar to some other networks described in this paper, a deep residual network can include convolutional layers followed by fully connected layers, which can be configured and trained in combination to extract feature attributes. The layers of a deep residual network can be configured to learn a residual function relative to the layer input, rather than learning a no-reference function. The layers of a deep residual network can be explicitly allowed to fit a residual mapping, rather than relying on a few stacked layers to directly fit the desired underlying mapping; this residual mapping can be achieved through a feedforward neural network with shortcut connections (i.e., connections that skip one or more layers). Deep residual networks can be created by modifying a regular neural network into a residual learning network by inserting shortcut connections into a regular neural network architecture that includes convolutional layers.
[0416] In some implementations, the machine learning module may include support vector machines (SVM), artificial neural networks (ANN), decision tree-based expert learning systems, autoencoders, clustering machine learning algorithms or nearest neighbor (e.g., kNN) machine learning algorithms or combinations thereof, some of which will be described in further detail below.
[0417] Support Vector Machines (SVMs) can be supervised learning algorithms used for classification and regression analysis of entity classification data or process control. Given a set of training data examples (e.g., entity data or process data), each example labeled as belonging to a certain category, an SVM training algorithm can build a model that assigns new examples (e.g., data from new entities or processes) to the given categories.
[0418] Figure 8 An artificial neural network (ANN) 800 according to one embodiment is illustrated. According to various embodiments, the ANN 800 can be used particularly for classification or process control optimization.
[0419] ANN 800 can include any type of neural network module, such as feedforward neural networks, radial basis function networks, recurrent neural networks, or convolutional neural networks.
[0420] In implementations of ANN 800 for entity classification, ANN 800 can be used to map entity data to entity classification data. In implementations of ANN 800 for process optimization, ANN 800 can be used to determine a set or series of Pareto optimal process control parameter settings, for example, by observing or providing process monitoring data streams and / or entity classification data from one or more sensors, to adaptively control the process in real time. ANN 800 can include untrained ANNs, trained ANNs, pre-trained ANNs, and continuously updated ANNs (e.g., ANNs using training data that is continuously updated with real-time classification data or process control and monitoring data from a single local system, multiple local systems, or multiple geographically distributed systems).
[0421] ANN 800 can include (e.g.) and The interconnected nodes are organized into n node layers, where This represents a set of i nodes in input layer 802 (e.g., layer 1). This indicates one or more hidden layers 803 (e.g., layer 2 to layer 803). A set of j nodes in a layer), and This represents a set of k nodes in the final layer 804 (e.g., the nth layer). The input layer 802 can be configured to receive input data 801 (e.g., sensor data, image data, sound data, observation data, automatically retrieved data, manually entered data, etc.). The final layer 804 can be configured to provide result data 805.
[0422] One or more hidden layers 803 may exist, and the number of nodes j in the one or more hidden layers 803 may vary depending on the implementation. Therefore, ANN 800 may include any total number of layers (e.g., one or more hidden layers 803). One or more hidden layers 803 can be used as trainable feature extractors, which can allow input data 801 to be mapped to preferred result data 805.
[0423] Figure 9 A node 900 according to one embodiment is illustrated. Each layer of the neural network may include one or more nodes similar to node 900, for example, Figure 8 The nodes shown and Each node can resemble a biological neuron.
[0424] Node 900 can receive node input 901 directly from the input data of the ANN (e.g., input data 801) or from the output of one or more nodes in different layers or the same layer (e.g., ...). Using node input 901, node 900 can execute operation 903, although... Figure 9 It is described as a summation operation, but it will be readily understood by those skilled in the art that it may include a variety of other operations known in the art.
[0425] In some cases, node input 901 can be paired with one or more weights 902 (e.g., The weights are associated with each node, and these weights can represent weighting factors. For example, operation 903 can sum the product of each node input 901 and its associated weight 902 (e.g., ...). ).
[0426] The result of operation 903 can be offset using one or more biases 904 (e.g., bias b), which can be values or functions.
[0427] The output 906 of node 900 can be gated using an activation (or threshold) function 905 (e.g., function f), which can be a linear or nonlinear function. The activation function 905 can be, for example, the ReLU activation function, or other functions such as the saturated hyperbolic tangent function, identity function, binary step function, logic function, arctangent function, softsign function, parameterized modified linear unit function, exponential linear unit function, softPlus function, bending identity function, softExponential function, sine function, Sinc function, Gaussian function, or sigmoid function, or any combination thereof.
[0428] During the training phase, one or more sets of training data can be used to "teach" or "learn" the weights 902, biases 904, thresholds for activation functions 905, or other computational parameters of the neural network. For example, these parameters can be trained using input data from the training dataset along with gradient descent or backpropagation methods, such that the output values computed by the ANN (e.g., a set of predictive adjustments for classification or process control parameter settings) are consistent with the examples included in the training dataset. These parameters can, for example, be obtained from a backpropagation neural network training process, which may or may not be performed using the same hardware used for automatic classification or adaptive real-time deposition process control.
[0429] Decision tree-based expert systems can be supervised learning algorithms designed to solve entity classification or process control problems by applying a set of conditional rules (e.g., if-then rules). An expert system may include two subsystems: an inference engine and a knowledge base. The knowledge base may include a set of facts (e.g., entity data comprising a range of entities and a training dataset of relevant entity classification data provided, for example, by skilled operators, technicians, or inspectors) and derived rules (e.g., derived entity classification rules). The inference engine can then apply these rules to the input data of the current entity classification or process control problem to determine the entity classification or the next set of process control adjustments.
[0430] An autoencoder (sometimes also called an autoassociator or Diablo network) can be an ANN used to map input data (e.g., entity data or process data) to output values (e.g., entity classification or optimized process control parameters) in an unsupervised and efficient manner. Autoencoders can be used for dimensionality reduction, which is the process of reducing the number of random variables considered by deriving a set of principal component variables. Dimensionality reduction can be performed, for example, for the purpose of feature selection (e.g., selecting a subset of the original variables) or feature extraction (e.g., transforming data in a high-dimensional space to a space with fewer dimensions).
[0431] Figure 10 A method 1000 for training a machine learning module according to one embodiment is illustrated. Using method 1000, training data can be used to train a machine learning model for simultaneous or subsequent use.
[0432] At position 1001, a machine learning model including one or more machine learning algorithms can be provided.
[0433] At point 1002, training data may be provided. Training data may include one or more of the following: process simulation data, process characterization data, in-process or post-process inspection data (including inspection data provided by skilled operators and / or by any of a variety of automated inspection tools), or any combination of the above for past processes that are the same as or different from the current process. One or more sets of training data may be used to train a machine learning algorithm for object defect detection and classification. In some cases, the data types included in the training dataset may vary depending on the specific type of machine learning algorithm employed.
[0434] At point 1003, training data can be used to train a machine learning model. For example, training a model may include feeding training data into the machine learning model and modifying one or more parameters of the model until the model's output is the same (or substantially the same) as external validation data. Model training can generate one or more trained models. One or more trained models can be selected for further validation or deployment, which can be performed using validation data. The results produced by each trained model against the validation data input to the training model can be compared with the validation data to determine which model is the best. For example, the trained model that produces results that most closely match the validation data can be selected as the best model. The selected model can then be evaluated using test data. The selected models can also be sent for model deployment, where the best model can be sent to a processor for use in post-trained mode.
[0435] Figure 11 A method 1100 for analyzing input data using a machine learning module according to one embodiment is illustrated. The machine learning module described in method 1100 can, for example, enable automatic classification of entities or optimized process control.
[0436] At point 1101, a trained machine learning model can be provided to the machine learning module. This trained model may have already been trained by one or more other systems or methods, or may be undergoing ongoing or periodic training. The machine learning model can be pre-generated and trained to implement the module functionality described herein, and can then be used to execute one or more post-trained functions of the machine learning module.
[0437] For example, the provided trained machine learning model may be similar to ANN 800, including nodes similar to node 900, and may have been trained (or is undergoing continuous or periodic training) using a method similar to method 1000.
[0438] At point 1102, input data can be provided to the machine learning module to be fed into the machine learning model. The input data can come from or originate from various different sources similar to input data 801.
[0439] Providing input data at 1102 can also include removing noise from the data before feeding it to the machine learning algorithm. Examples of suitable data processing algorithms for removing noise from input data include, in particular, signal averaging algorithms, smoothing filtering algorithms, Kalman filtering algorithms, nonlinear filtering algorithms, total variation minimization algorithms, or any combination thereof.
[0440] Providing input data at point 1102 can also include subtracting a reference dataset from the input data to increase the contrast between aspects of interest and those of no interest in the entity or process, thereby aiding in classification or process control optimization. For example, the reference dataset can include input data that represents a real or constructed ideal example of the entity or process. If the entity is observed using an image sensor or machine vision system, the reference dataset can include images of the ideal entity or a collection of images (e.g., representing different views).
[0441] At point 1103, the machine learning module can use a trained machine learning model to process the input data to produce results for the machine learning module. These results may include, for example, entity classification or one or more optimized process control parameters.
[0442] The following terms may provide additional context for this disclosure, but should in no way be considered restrictive.
[0443] Clause 1. A method comprising: receiving a set of mechanisms at a processor of one of a plurality of nodes, wherein each mechanism is defined at least by a provider action and a feasible contract; using the processor, defining a receiver loss function based on the set of mechanisms; using the processor, determining the mechanism in the set of mechanisms that minimizes the receiver loss function; using the processor, generating a smart contract based on the mechanism; using the processor, deploying the smart contract to a blockchain maintained across the plurality of nodes connected via a computer network by broadcasting the smart contract to the plurality of nodes via a network interface that communicates electronically with the processor and a computer network; receiving at the processor a transaction address corresponding to the smart contract from the blockchain; and using the processor, storing the transaction address in a database located on an electronic storage device at the node and communicating electronically with the processor.
[0444] Clause 2. The method according to Clause 1, the method further comprising: the processor sending a data request to the blockchain in response to the transaction address to verify the smart contract.
[0445] Clause 3. The method according to one or more of Clauses 1 to 2, the method further comprising receiving oracle data at the processor, wherein each mechanism is further defined by the oracle data.
[0446] Clause 4. The method according to Clause 3, wherein the oracle data is generated by: the processor sending a hint to a trained machine learning model; and receiving the output of the trained machine learning model at the processor as the oracle data.
[0447] Clause 5. The method according to Clause 4, wherein receiving the output of the trained machine learning model as the oracle data comprises: using the processor to perform data operations on the raw output data of the trained machine learning model to generate the oracle data.
[0448] Clause 6. The method described in one or more of Clauses 3 to 5, wherein the oracle data is configured by prompts received from the provider node.
[0449] Clause 7. The method described in one or more of Clauses 3 to 6, wherein the oracle data is configured by a hint received from the receiving node.
[0450] Clause 8. The method described in one or more of Clauses 1 to 7, further comprising: using the processor to deploy transactions of the tokens to the blockchain.
[0451] Clause 9. The method described in Clause 8, wherein the token is a non-fungible token.
[0452] Clause 10. The method described in Clause 9, wherein the non-fungible token is a non-destroyable non-fungible token.
[0453] Clause 11. The method described in one or more of Clauses 1 to 10, wherein each mechanism is further defined by provider type, provider signal and provider remuneration.
[0454] Clause 12. The method described in one or more of Clauses 1 to 11, wherein each feasible contract is defined by the provider signal and the provider action of the mechanism.
[0455] Clause 13. The method described in accordance with Clause 12, wherein each viable contract is further limited by remuneration from the provider.
[0456] Clause 14. The method described in Clause 13, wherein the provider remuneration is selected by the user from a set of possible provider remunerations.
[0457] Clause 15. The method according to Clause 14, wherein the set of possible provider remuneration is selected based on the provider signal.
[0458] Clause 16. The method described in one or more of Clauses 1 to 15, wherein each viable contract is subject to one or more of the incentive compatibility constraint, individual rationality constraint, or solvency constraint.
[0459] Clause 17. The method according to one or more of Clauses 1 to 16, the method further comprising: using the processor to determine provider verification data; and using the processor to deploy the provider verification data to the blockchain.
[0460] Clause 18. The method described in Clause 17, wherein the provider verification data includes the provider reputation.
[0461] Clause 19. The method described in one or more of Clauses 17 to 18, wherein the provider verification data includes certificate ratings.
[0462] Clause 20. The method according to one or more of Clauses 1 to 19, the method further comprising: using the processor to retrieve provider verification data from the blockchain, wherein each mechanism is further defined by the provider verification data.
[0463] Clause 21. A system comprising: a processor of one of a plurality of nodes; a network interface electronically communicating with the processor and a computer network; and an electronic storage device electronically communicating with the processor, the electronic storage device storing a database; wherein the processor is configured to: receive a set of mechanisms, wherein each mechanism is defined at least by a provider action and a feasible contract; define a receiver loss function based on the set of mechanisms; determine the mechanism in the set of mechanisms that minimizes the receiver loss function; generate a smart contract based on the mechanism; deploy the smart contract to a blockchain maintained by the plurality of nodes connected via a computer network by broadcasting the smart contract to the plurality of nodes via the network interface; receive a transaction address corresponding to the smart contract from the blockchain; and store the transaction address in the database of the electronic storage device.
[0464] Clause 22. The system according to Clause 21, wherein the processor is further configured to send a data request to the blockchain in response to the transaction address to verify the smart contract.
[0465] Clause 23. A system according to one or more of Clauses 21 to 24, wherein the processor is further configured to receive oracle data, wherein each mechanism is further defined by the oracle data.
[0466] Clause 24. The system according to Clause 23, wherein the oracle data is generated by: the processor sending a hint to a trained machine learning model; and receiving the output of the trained machine learning model at the processor as the oracle data.
[0467] Clause 25. The system according to Clause 24, wherein receiving the output of the trained machine learning model as the oracle data comprises: performing data operations on the raw output data of the trained machine learning model to generate the oracle data.
[0468] Clause 26. The system as described in Clause 23, wherein the oracle data is configured by prompts received from the provider node.
[0469] Clause 27. A system according to one or more of Clauses 23 to 26, wherein the oracle data is configured by prompts received from the receiving node.
[0470] Clause 28. A system according to one or more of Clauses 21 to 27, wherein the processor is further configured to deploy transactions of tokens to the blockchain.
[0471] Clause 29. The system described in Clause 28, wherein the tokens are non-fungible tokens.
[0472] Clause 30. The system described in Clause 29, wherein the nonfungible token is a non-destructible nonfungible token.
[0473] Clause 31. A system pursuant to one or more of Clauses 21 to 30, wherein each mechanism is further defined by provider type, provider signal and provider remuneration.
[0474] Clause 32. A system pursuant to one or more of Clauses 21 to 32, wherein each viable contract is defined by a provider signal and the provider action of the mechanism.
[0475] Clause 33. The system as described in Clause 32, wherein each viable contract is further limited by remuneration from the provider.
[0476] Clause 34. The system described in Clause 33, wherein the provider remuneration is selected by the user from a set of possible provider remunerations.
[0477] Clause 35. The system described in Clause 34, wherein the set of possible provider remunerations is selected based on the provider signal.
[0478] Clause 36. A system pursuant to one or more of Clauses 21 to 35, wherein each viable contract is subject to one or more of incentive compatibility constraints, individual rationality constraints, or solvency constraints.
[0479] Clause 37. A system according to one or more of Clauses 21 to 36, wherein the processor is further configured to: determine provider verification data; and deploy the provider verification data to the blockchain.
[0480] Clause 38. The system described in Clause 37, wherein the provider verification data includes provider reputation.
[0481] Clause 39. A system pursuant to one or more of Clauses 37 to 38, wherein the provider verification data includes certificate ratings.
[0482] Clause 40. A system according to one or more of Clauses 21 to 39, wherein the processor is further configured to retrieve provider verification data from the blockchain, wherein each mechanism is further defined by the provider verification data.
[0483] Clause 41. A tangible, non-transitory computer-readable medium having instructions, which, when implemented by a processor of one of a plurality of nodes, cause the processor to perform a method, the method comprising: receiving a set of mechanisms, wherein each mechanism is defined at least by a provider action and a feasible contract; defining a receiver loss function based on the set of mechanisms; determining the mechanism among the set of mechanisms that minimizes the receiver loss function; generating a smart contract based on the mechanism; deploying the smart contract to a blockchain maintained by the plurality of nodes connected via the computer network by broadcasting the smart contract to the plurality of nodes via a network interface that communicates electronically with the processor and the computer network; receiving a transaction address corresponding to the smart contract from the blockchain; and storing the transaction address in a database located on an electronic storage device at the node and in electronic communication with the processor.
[0484] Clause 42. The tangible, non-transitory, computer-readable medium as described in Clause 41, wherein the method further includes sending a data request to the blockchain in response to the transaction address to verify the smart contract.
[0485] Clause 43. A tangible, non-transitory, computer-readable medium according to one or more of Clauses 41 to 42, wherein the method further includes receiving oracle data, wherein each mechanism is further defined by the oracle data.
[0486] Clause 44. The tangible, non-transitory, computer-readable medium as described in Clause 43, wherein the oracle data is generated by: sending a hint from the processor to a trained machine learning model; and receiving the output of the trained machine learning model at the processor as the oracle data.
[0487] Clause 45. The tangible, non-transitory, computer-readable medium as described in Clause 44, wherein receiving the output of the trained machine learning model as the oracle data comprises: performing data operations on the raw output data of the trained machine learning model to generate the oracle data.
[0488] Clause 46. A tangible, non-transitory, computer-readable medium as described in Clause 43, wherein the oracle data is configured by prompts received from the provider node.
[0489] Clause 47. A tangible, non-transitory, computer-readable medium according to one or more of Clauses 43 to 46, wherein the oracle data is configured by prompts received from the receiving node.
[0490] Clause 48. A tangible, non-transitory, computer-readable medium as described in one or more of Clauses 41 to 47, wherein the method further includes deploying transactions of tokens to the blockchain.
[0491] Clause 49. A tangible, non-transitory, computer-readable medium as described in Clause 48, wherein the token is a non-fungible token.
[0492] Clause 50. A tangible, non-transitory, computer-readable medium as described in Clause 49, wherein the non-fungible token is a non-destructible non-fungible token.
[0493] Clause 51. A tangible, non-transitory, computer-readable medium as described in one or more of Clauses 41 to 50, wherein each mechanism is further defined by provider type, provider signal, and provider remuneration.
[0494] Clause 52. A tangible, non-transitory, computer-readable medium pursuant to one or more of Clauses 41 to 51, wherein each viable contract is defined by a provider signal and the provider action of the mechanism.
[0495] Clause 53. A tangible, non-transitory, computer-readable medium as described in Clause 52, wherein each viable contract is further limited by remuneration from the provider.
[0496] Clause 54. A tangible, non-transitory, computer-readable medium as described in Clause 53, wherein the provider’s compensation is selected by the user from a set of possible provider compensations.
[0497] Clause 55. A tangible, non-transitory, computer-readable medium as described in Clause 54, wherein the set of possible provider rewards is selected based on the provider signal.
[0498] Clause 56. A tangible, non-transitory, computer-readable medium pursuant to one or more of Clauses 41 to 55, wherein each viable contract is subject to one or more of incentive compatibility constraints, individual rationality constraints, or solvency constraints.
[0499] Clause 57. A tangible, non-transitory, computer-readable medium according to one or more of Clauses 41 to 56, wherein the method further comprises: determining provider verification data; and deploying the provider verification data to the blockchain.
[0500] Clause 58. A tangible, non-transitory, computer-readable medium as described in Clause 57, wherein the provider verification data includes the provider's reputation.
[0501] Clause 59. A tangible, non-transitory, computer-readable medium as described in one or more of Clauses 57 to 58, wherein the provider verification data includes certificate ratings.
[0502] Clause 60. A tangible, non-transitory, computer-readable medium according to one or more of Clauses 41 to 59, wherein the method further includes retrieving provider verification data from the blockchain, wherein each mechanism is further defined by the provider verification data.
[0503] This invention is limited only by the appended claims. Variations, features, advantages, embodiments, constructions, arrangements, terms, materials, dimensions, examples, illustrations, depictions, and examples constituting the above description and drawings illustrate some possible embodiments of the invention, but do not limit the invention. Not every embodiment of the invention achieves or possesses every advantage, object, or feature pointed out herein, and therefore, those skilled in the art can make various additions, changes, modifications, or omissions without departing from the scope or spirit of the invention or its legal equivalents.
[0504] Unless otherwise defined, all ranges include the specified limit, its order of magnitude, and substantially all values and ranges within it. Unless otherwise stated, each use of "and" forms an inclusive list that includes at least the connected elements, and each use of "or" forms an inclusive list that includes at least one of the connected elements. Unless otherwise stated, singular uses (e.g., "a," "an," or "the") include their plural forms.
[0505] The order of listing in the claims does not imply a temporal or sequential relationship unless the express wording of the claim unavoidably requires such a relationship. Unless a claim specifies "means for..." or "steps for...", it shall not be construed as invoking section 212(f) of Title 35 of the United States Code.
Claims
1. A method, the method comprising: A set of mechanisms is received at the processor of one of the multiple nodes, wherein each mechanism is defined at least by a provider action and a feasible contract; Using the processor, the receiver loss function is defined based on the set of mechanisms; Using the processor, determine the mechanism in the set of mechanisms that minimizes the receiver's loss function; Using the processor, smart contracts are generated based on the mechanism; Using the processor, the smart contract is broadcast to multiple nodes via a network interface that communicates electronically with the processor and the computer network, thereby deploying the smart contract to a blockchain maintained by the multiple nodes connected via the computer network; The processor receives the transaction address corresponding to the smart contract from the blockchain. as well as The processor is used to store the transaction address in a database located on the electronic storage device of the node and in electronic communication with the processor.
2. The method according to claim 1, further comprising: The processor responds to the transaction address by sending a data request to the blockchain to verify the smart contract.
3. The method according to claim 1, further comprising: Oracle data is received at the processor, wherein each mechanism is further defined by the oracle data.
4. The method of claim 3, wherein the oracle data is generated by: the processor sending a hint to a trained machine learning model; and receiving the output of the trained machine learning model at the processor as the oracle data.
5. The method of claim 4, wherein receiving the output of the trained machine learning model as the oracle data comprises: Using the processor, data operations are performed on the raw output data of the trained machine learning model to generate the oracle data.
6. The method of claim 3, wherein the oracle data is configured by prompts received from the provider node.
7. The method of claim 3, wherein the oracle data is configured by a prompt received from the receiving node.
8. The method according to claim 1, further comprising: Using the processor, token transactions are deployed to the blockchain.
9. The method of claim 8, wherein the token is a non-fungible token.
10. The method of claim 9, wherein the non-fungible token is a non-destroyable non-fungible token.
11. The method of claim 1, wherein each mechanism is further defined by provider type, provider signal, and provider reward.
12. The method of claim 1, wherein each viable contract is defined by a provider signal and the provider action of the mechanism.
13. The method of claim 12, wherein each feasible contract is further defined by provider remuneration.
14. The method of claim 13, wherein the provider compensation is selected by the user from a set of possible provider compensations.
15. The method of claim 14, wherein the set of possible provider rewards is selected based on the provider signal.
16. The method of claim 1, wherein each feasible contract is subject to one or more of incentive compatibility constraints, individual rationality constraints, or solvency constraints.
17. The method according to claim 1, further comprising: Using the processor, the provider's verification data is determined; as well as Using the processor, the provider verification data is deployed to the blockchain.
18. The method of claim 17, wherein the provider verification data includes provider reputation.
19. The method of claim 17, wherein the provider verification data includes certificate ratings.
20. The method according to claim 1, further comprising: Using the processor, provider verification data is retrieved from the blockchain, wherein each mechanism is further defined by the provider verification data.
21. A system comprising: The processor of one of a set of multiple nodes; A network interface that communicates electronically with the processor and the computer network; and An electronic storage device that communicates electronically with the processor, and the electronic storage device stores a database. The processor is configured as follows: Receive a set of mechanisms, each of which is defined by at least provider actions and feasible contracts; The receiver loss function is defined based on the aforementioned set of mechanisms; Identify the mechanism in the set of mechanisms that minimizes the receiver's loss function; Smart contracts are generated based on the aforementioned mechanism; The smart contract is deployed to a blockchain maintained by the multiple nodes connected via a computer network by broadcasting the smart contract to multiple nodes via the network interface. Receive the transaction address corresponding to the smart contract from the blockchain; as well as The transaction address is stored in the database of the electronic storage device.
22. The system of claim 21, wherein the processor is further configured to send a data request to the blockchain in response to the transaction address to verify the smart contract.
23. The system of claim 21, wherein the processor is further configured to receive oracle data, wherein each mechanism is further defined by the oracle data.
24. The system of claim 23, wherein the oracle data is generated in the following manner: The processor sends prompts to the trained machine learning model; and The processor receives the output of the trained machine learning model as the oracle data.
25. The system of claim 24, wherein receiving the output of the trained machine learning model as the oracle data comprises: Data operations are performed on the raw output data of the trained machine learning model to generate the oracle data.
26. The system of claim 23, wherein the oracle data is configured by prompts received from the provider node.
27. The system of claim 23, wherein the oracle data is configured by prompts received from the receiving node.
28. The system of claim 21, wherein the processor is further configured to deploy token transactions to the blockchain.
29. The system of claim 28, wherein the token is a non-fungible token.
30. The system of claim 29, wherein the nonfungible token is a non-destructible nonfungible token.
31. The system of claim 21, wherein each mechanism is further defined by provider type, provider signal, and provider reward.
32. The system of claim 21, wherein each viable contract is defined by a provider signal and the provider action of the mechanism.
33. The system of claim 32, wherein each feasible contract is further limited by provider remuneration.
34. The system of claim 33, wherein the provider compensation is selected by the user from a set of possible provider compensations.
35. The system of claim 34, wherein the set of possible provider rewards is selected based on the provider signal.
36. The system of claim 21, wherein each feasible contract is subject to one or more of incentive compatibility constraints, individual rationality constraints, or solvency constraints.
37. The system of claim 21, wherein the processor is further configured to: Determine the provider's verification data; and The provider's verification data is deployed to the blockchain.
38. The system of claim 37, wherein the provider verification data includes provider reputation.
39. The system of claim 37, wherein the provider verification data includes certificate ratings.
40. The system of claim 21, wherein the processor is further configured to retrieve provider verification data from the blockchain, wherein each mechanism is further defined by the provider verification data.
41. A tangible, non-transitory computer-readable medium having instructions that, when implemented by a processor of one of a plurality of nodes, cause the processor to perform a method, the method comprising: Receive a set of mechanisms, each of which is defined by at least provider actions and feasible contracts; The receiver loss function is defined based on the aforementioned set of mechanisms; Identify the mechanism in the set of mechanisms that minimizes the receiver's loss function; Smart contracts are generated based on the aforementioned mechanism; The smart contract is deployed to a blockchain maintained by the multiple nodes connected via the computer network by broadcasting the smart contract to multiple nodes via a network interface that communicates electronically with the processor and computer network. Receive the transaction address corresponding to the smart contract from the blockchain; as well as The transaction address is stored in a database located on the electronic storage device of the node and communicates electronically with the processor.
42. The tangible, non-transitory, computer-readable medium of claim 41, wherein the method further comprises sending a data request to the blockchain in response to the transaction address to verify the smart contract.
43. The tangible, non-transitory, computer-readable medium of claim 41, wherein the method further comprises receiving oracle data, wherein each mechanism is further defined by the oracle data.
44. The tangible, non-transitory, computer-readable medium of claim 43, wherein the oracle data is generated in the following manner: The processor sends prompts to the trained machine learning model; and The processor receives the output of the trained machine learning model as the oracle data.
45. The tangible, non-transitory computer-readable medium of claim 44, wherein receiving the output of the trained machine learning model as the oracle data comprises: Data operations are performed on the raw output data of the trained machine learning model to generate the oracle data.
46. The tangible, non-transitory, computer-readable medium of claim 43, wherein the oracle data is configured by prompts received from the provider node.
47. The tangible, non-transitory, computer-readable medium of claim 43, wherein the oracle data is configured by prompts received from the receiving node.
48. The tangible, non-transitory, computer-readable medium of claim 41, wherein the method further comprises deploying token transactions to the blockchain.
49. The tangible, non-transitory, computer-readable medium of claim 48, wherein the token is a non-fungible token.
50. The tangible, non-transitory, computer-readable medium of claim 49, wherein the non-fungible token is a non-destructible non-fungible token.
51. The tangible, non-transitory computer-readable medium of claim 41, wherein each mechanism is further defined by a provider type, a provider signal, and a provider reward.
52. The tangible, non-transitory, computer-readable medium of claim 41, wherein each viable contract is defined by a provider signal and the provider action of the mechanism.
53. The tangible, non-transitory, computer-readable medium of claim 52, wherein each viable contract is further defined by remuneration from the provider.
54. The tangible, non-transitory computer-readable medium of claim 53, wherein the provider compensation is selected by the user from a set of possible provider compensations.
55. The tangible, non-transitory computer-readable medium of claim 54, wherein the set of possible provider rewards is selected based on the provider signal.
56. The tangible, non-transitory, computer-readable medium of claim 41, wherein each feasible contract is subject to one or more of incentive compatibility constraints, individual rationality constraints, or solvency constraints.
57. The tangible, non-transitory computer-readable medium of claim 41, wherein the method further comprises: Determine the provider to verify the data; as well as The provider's verification data is deployed to the blockchain.
58. The tangible, non-transitory computer-readable medium of claim 57, wherein the provider verification data includes provider reputation.
59. The tangible, non-transitory, computer-readable medium of claim 57, wherein the provider verification data includes certificate ratings.
60. The tangible, non-transitory computer-readable medium of claim 41, wherein the method further comprises retrieving provider verification data from the blockchain, wherein each mechanism is further defined by the provider verification data.
