A decentralized AI algorithm and system

Abstract

The present invention proposes a decentralized AI computing method and system, which aims to optimize AI model training, protect user privacy and improve resource utilization efficiency through decentralized networks and blockchain technology. The method includes locally executing AI training on multiple network nodes, generating model updates, and uploading them to the decentralized network through a secure protocol. Smart contracts automatically aggregate these updates, generate and distribute aggregated model updates. The system consists of nodes with data processing and encryption functions and a decentralized network including model aggregation and incentive distribution. The present invention realizes secure data transmission, automatic incentive distribution, and encourages user participation, promoting the development of decentralized AI technology.

Description

Technical Field

[0001] The present invention relates to artificial intelligence (AI) and blockchain technology, and in particular to a decentralized AI computing method and system, which aims to optimize the training process of AI models, enhance user privacy protection, improve resource utilization efficiency, and realize user incentives and community governance through smart contracts. Background Art

[0002] In the modern technological environment, artificial intelligence (AI) has become a key force driving the development of all industries. The core of AI technology relies on the processing power of big data and powerful computing resources to automate complex tasks from image recognition to natural language processing. However, with the rapid growth of AI applications, the existing centralized AI training system faces several key challenges.

[0003] First, the centralized data processing and storage model has caused serious privacy protection issues. In this model, user data needs to be uploaded to the central server for processing, which not only increases the risk of illegal access and abuse of data, but also greatly weakens the user's control over their own data. In addition, once the central server is attacked, it will lead to large-scale data leakage, posing a major threat to user privacy security.

[0004] Secondly, AI training systems that rely on centralized computing resources often face problems of uneven resource allocation and low utilization. Central servers need to process a large number of data requests from all over the world, which can easily cause resource bottlenecks and affect training efficiency and speed. In addition, computing resources under a centralized architecture are often concentrated in the hands of a few providers, resulting in resource monopoly, exacerbating cost increases, and limiting the participation opportunities of small businesses and individual developers.

[0005] Finally, the centralized decision-making and resource control mechanism lacks transparency and fairness. In this model, all decisions and resource allocation are controlled by the central server, and it is difficult for users and small developers to participate in the selection and optimization of AI models. This limits the innovation and diversified development of AI technology and weakens the consensus and trust of society on the development direction of AI technology.

[0006] Therefore, finding a new AI training system that can protect user data privacy, efficiently utilize decentralized computing resources, and achieve transparency in resource control and decision-making processes has become a key requirement for promoting the healthy development of AI technology. This requires a new technical architecture to address the challenges faced by traditional centralized AI systems and promote the development of AI technology in a safer, more efficient and fairer direction. Summary of the invention

[0007] This invention aims to overcome the problems of privacy, resource utilization efficiency, and governance transparency in traditional centralized AI training systems by proposing a decentralized AI computing method and system implementation. The core idea is to use decentralized technology - especially blockchain and smart contracts - to build a safe, efficient, and fair AI training environment. The following are the key contents of this invention:

[0008] Decentralized AI training framework

[0009] The present invention builds a decentralized network framework, so that each node (including personal computers, smart phones, servers, etc.) distributed in the network can use its own idle computing resources to participate in the training of AI models. This distributed computing model can not only significantly improve the utilization of computing resources, but also avoid the privacy leakage risks brought about by centralized storage and processing of data.

[0010] Application of blockchain technology

[0011] The present invention uses blockchain technology to ensure data security and transaction transparency. The contributions of all participating nodes (such as data provision, computing power contribution, etc.) and the updates of AI models are recorded and verified through blockchain. This not only ensures the data immutability and traceability of the system, but also realizes the full transparency of the training process, providing a trust basis for users.

[0012] Smart Governance of Smart Contracts

[0013] The present invention uses smart contracts to automatically manage user incentives and community governance. Smart contracts automatically evaluate the contribution of each participating node according to preset rules, and allocate corresponding incentives (such as tokens, points, etc.) accordingly. In addition, smart contracts can also support community members to vote on key decisions such as the selection of AI models and system updates, realizing decentralized democratic governance.

[0014] Enhanced protection of user privacy

[0015] In this invention, user data will be encrypted locally before participating in AI training to ensure that sensitive information will not be leaked during the training process. At the same time, by adopting advanced encryption communication protocols, the security of user data and model update information is guaranteed during the process of uploading to the decentralized network.

[0016] In general, the present invention provides a new AI computing method and system by combining decentralized technology, blockchain and smart contracts, aiming to solve the problems existing in the existing centralized AI training system and provide a safer, more efficient and fair solution for the development of AI technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0018] Figure 1 A flowchart of a decentralized AI computing method;

[0019] Figure 2 It is a structural diagram of a decentralized AI computing system;

[0020] Figure 3 This is an example diagram of decentralized AI training and services applied to medical systems.

[0021] Technical Solution

[0022] like Figure 1 , Figure 2 The present invention proposes an innovative decentralized AI computing method and its system implementation, aiming to improve the efficiency, security and fairness of AI model training through the application of decentralized computing resources and blockchain technology. The following describes in detail the method steps and system composition of the present invention.

[0023] Steps of decentralized AI computing method:

[0024] Deployment and execution of AI model training tasks: This method first deploys AI model training tasks on multiple nodes of a decentralized network. These nodes include but are not limited to personal computers, smartphones, and cloud servers. Each node uses its local data to train the AI ​​model and independently generates model updates. This not only makes full use of decentralized computing resources, but also ensures the privacy and security of training data.

[0025] Upload and share model updates: After training is completed, each node uploads the model update to the shared data pool in the decentralized network through a secure decentralized network communication protocol. This protocol ensures encryption protection of data during transmission and ensures the security of model update information.

[0026] Smart contract processing of model updates: Smart contracts in the network are automatically activated to collect model updates from the shared data pool for verification and aggregation. Through the algorithm of smart contracts, the authenticity and validity of model updates can be ensured, while generating global aggregate model updates.

[0027] Distribution of aggregated model updates: Finally, the aggregated model updates are distributed to all participating computing nodes. Each node synchronously updates the local AI model based on the aggregated update information to improve the overall performance and accuracy of the model.

[0028] The composition of the decentralized AI computing system:

[0029] Compute nodes: The basic building blocks of the system, equipped with necessary data processing, storage, and encryption processing units. Compute nodes are responsible for performing specific AI model training tasks, generating model updates, and securely uploading these updates to the decentralized network.

[0030] Decentralized network: As the core of the system, the decentralized network includes data verification units, model aggregation units, and incentive allocation units. Smart contracts play a role in automatically processing model update information in the network, ensuring the operating efficiency and security of the entire system.

[0031] Decentralized Communication Protocol: This protocol supports secure data exchange between computing nodes and decentralized networks, ensuring the security and privacy of model update information during transmission.

[0032] Through the above method and system implementation, the present invention effectively solves the problems existing in the traditional centralized AI training system, such as data privacy leakage risk, low utilization of computing resources and lack of transparency, and provides a new solution for the development of AI technology. By utilizing decentralized technology and blockchain technology, the present invention not only protects the privacy security of users, but also realizes the efficient utilization and fair allocation of computing resources. Example

[0033] like Figure 3 In order to specifically illustrate the application of the present invention, this embodiment describes an AI training project based on medical image recognition, and implements the execution and management of the project by utilizing the decentralized AI computing method and system proposed in the present invention.

[0034] Implementation Background

[0035] Image recognition technology is an important branch of AI and is widely used in security monitoring, medical diagnosis, autonomous driving and other fields. With the development of technology, the demand for training image recognition models continues to grow, which requires more data and computing resources. Traditional centralized training methods are limited by privacy and resource utilization issues. Therefore, the decentralized AI computing method and system of the present invention are used to train image recognition models, which can not only protect data privacy, but also improve resource utilization efficiency.

[0036] Implementation steps

[0037] Project launch: Launch an AI training project based on medical image recognition and invite users of the medical alliance to participate. Each participating medical institution joins the decentralized network as an independent computing node.

[0038] Model training: With the user's authorization, each computing node uses the local imaging data stored in the hospital to train the specified AI model. The training process is completely performed locally and does not involve external data transmission, thus ensuring the privacy and security of user data.

[0039] Model update upload: After training is completed, each computing node generates model updates and securely uploads these model updates to the shared data pool in the decentralized network through a decentralized communication protocol.

[0040] Smart Contract Processing: Pre-deployed smart contracts in the decentralized network are automatically activated to collect, verify, and aggregate model updates in the shared data pool to generate global aggregate model updates.

[0041] Model update distribution: The aggregated model update is distributed to all participating computing nodes, and each node updates the local AI model accordingly, completing a training cycle.

[0042] User incentives: Participating hospitals automatically receive incentives through smart contracts based on their data and computing power contributions. Incentives can be points or other forms, designed to encourage hospitals to actively participate in training programs.

[0043] Implementation Effect

[0044] Through this embodiment, not only the efficient training of the image recognition model is achieved, but also the data privacy security of the participating hospitals is ensured. The decentralized computing resource utilization and the automated management of smart contracts have greatly improved the transparency and fairness of the training projects. In addition, the introduction of the incentive mechanism has effectively stimulated the enthusiasm of users to participate and promoted the construction and development of the decentralized AI ecosystem.

[0045] In summary, this embodiment successfully demonstrated the feasibility and advantages of the present invention in practical applications, and provided a new path for the further development of AI technology.

[0046] Technical Effects

[0047] This invention combines decentralized computing, blockchain technology and smart contracts to propose a new AI computing method and system, aiming to solve the problems of privacy protection, low resource utilization efficiency and insufficient governance transparency in traditional centralized AI training systems. The implementation of this invention brings the following significant technical effects:

[0048] Effective protection of user privacy

[0049] By using a decentralized computing framework, the present invention realizes local processing of data, avoiding the risk of uploading sensitive data to a central server. At the same time, encryption technology and decentralized communication protocols are used to ensure the security of data during uploading and sharing. These measures effectively protect the privacy of users and reduce the risk of data leakage and abuse.

[0050] Improved resource utilization and transparency

[0051] By dynamically integrating the distributed computing resources in the network, the present invention significantly improves the resource utilization of AI training. Decentralized computing nodes can participate in or exit AI training tasks according to actual conditions, thereby achieving flexible scheduling and efficient utilization of resources. In addition, all model training tasks, model update processes and their results are recorded and verified through blockchain technology, ensuring the transparency and traceability of the training process.

[0052] Implementation of automatic incentive allocation and community governance

[0053] The present invention automatically manages user incentives and community governance through smart contracts, providing a fair and transparent incentive mechanism for user contributions (including data provision and computing power contribution). Users can obtain corresponding incentives, such as digital currency or points, based on their contributions, which motivates users to participate in AI training tasks. At the same time, smart contracts also support community members to vote on key decisions (such as AI model selection, system updates, etc.), realizing decentralized democratic governance.

[0054] Promote the development of decentralized AI technology

[0055] This invention provides a feasible implementation path for decentralized AI technology, which not only solves the core problems in the existing technology, but also promotes technological innovation and ecosystem construction through incentive mechanisms and community governance models. This helps to break the monopoly of traditional AI technology development and application.

[0056] In summary, through technological innovation, the present invention can not only improve the efficiency and economy of AI training, but more importantly, it will promote the fairness and sustainability of the development of AI technology and lay the foundation for building a more open, secure and fair AI ecosystem.

[0057] The above describes the preferred embodiments of the present invention. Of course, the implementation of the present invention is not limited to these, but also includes more variants and modifications, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A decentralized AI computing method, characterized in that: The specific implementation steps are as follows: Step 1: Register multiple computing nodes on the decentralized network through a distributed network registration mechanism, including but not limited to smartphones, personal computers, and cloud servers, etc. Each node is equipped with the necessary data processing and storage capabilities; Step 2: On the registered computing nodes, according to the predefined AI model training framework, use the local data on each node to perform model training tasks, and generate model update information based on the training results; Step 3: Each computing node uploads its model update information to the shared data pool in the decentralized network through a secure decentralized communication protocol, while ensuring that the data is encrypted for privacy protection; Step 4: The smart contract set in the decentralized network is automatically activated to collect, verify and aggregate the model update information in the shared data pool to generate a global aggregate model update; Step 5: Distribute the aggregated model update to all participating computing nodes through a decentralized network. Each node updates the local AI model accordingly to complete a training cycle.

2. Claim 2: The decentralized AI computing method according to claim 1, characterized in that: The smart contract further has the ability to evaluate the model update quality and contribution of each computing node, and allocate corresponding points to the nodes as incentives based on the evaluation results. The incentive mechanism is designed based on open and transparent rules to ensure fairness and effectiveness of incentives.

3. Claim 3: The decentralized AI computing method according to claim 1 or 2, characterized in that: The secure decentralized communication protocol uses advanced encryption technology, including but not limited to asymmetric encryption and zero-knowledge proof, to ensure the secure transmission and verification of model update information without exposing the original data content.

4. Claim 4: A decentralized AI computing system, characterized in that: include: Multiple computing nodes, each equipped with a data processing unit, a storage unit, an encryption processing unit, and a network communication unit, for performing local training of AI models, generating model update information, and supporting the upload and download of encrypted information; A decentralized network that includes a data verification unit, a model aggregation unit, and an incentive allocation unit. The network processes uploaded model update information through smart contracts and automatically allocates incentives based on the contribution of nodes; A decentralized communication protocol that supports secure data exchange between computing nodes and decentralized networks, using encryption technology to ensure the privacy and security of data transmission.

5. Claim 5: The decentralized AI computing system according to claim 4, characterized in that: The data processing unit is further integrated with high-performance computing capabilities and AI training optimization algorithms, and can automatically adjust the allocation of computing resources according to different AI tasks to optimize the efficiency and effectiveness of local training.

6. Claim 6: The decentralized AI computing system according to claim 4 or 5, characterized in that: The incentive allocation unit uses blockchain technology to record and execute the incentive allocation process, ensuring that the incentive allocation process is open, transparent, and cannot be tampered with, while supporting flexible configuration of multiple incentive forms, including but not limited to digital currency and points system.

Images