30 results about "Chain model" patented technology

A method, apparatus, and system for performing a task using a chain of thought model. An initiation prompt for execution by at least one machine learning model is provided, including an objective and instructions to define a plurality of agents for achieving the objective and an arrangement of the plurality of agents. A response to the initiation prompt is received defining the plurality of agents and the arrangement of the plurality of agents. The arrangement comprises a hypergraph in which a plurality of nodes represents the plurality of agents, respectively, and a plurality of edges represents data connections between the plurality of agents. A first agent and a second agent are instantiated based on the response. The first agent is represented by a first node of the plurality of nodes in the hypergraph. The second agent is represented by a second node. An output of the second agent is obtained.

The invention discloses a search intention recognition method and device, equipment and a storage medium, and relates to the technical field of reinforcement learning, and the method comprises the following steps: generating training data based on an original corpus; performing supervision fine tuning processing on the lightweight language model based on the training data to obtain a fine tuning trained lightweight language model; performing strategy optimization on the lightweight language model subjected to fine tuning training through an improved GRPO reinforcement learning algorithm to obtain a lightweight language model subjected to reinforcement training; constructing a double-model hybrid architecture based on a short thinking chain model and a thinking chain-free model in the lightweight language model subjected to enhanced training; and outputting a final sequence of the related documents of the search terms of the user through the double-model hybrid architecture. According to the technical scheme of lightweight language model fine tuning, reinforcement learning optimization and double-model hybrid reasoning, the target search scene recognition accuracy is improved, the real-time response speed is high, and training time consumption and reasoning resource requirements are reduced.

The invention relates to the technical field of encrypted communication, and discloses an intelligent analysis method and system based on multi-source data, and the method comprises the steps: obtaining multi-dimensional metadata of an encrypted communication flow, and generating a feature vector set; calculating a state transition probability of a communication session by using a Markov chain model, dividing the communication session according to a time window and mapping the communication session to a discrete state, and generating a session state evolution sequence; analyzing a session state evolution sequence based on a hierarchical topic model, mapping the state sequence to a behavior topic through potential Dirichlet allocation, mapping the behavior topic to an intention category through a hierarchical Dirichlet process, and outputting a communication intention category and a confidence score; according to the method, the limitation that static feature analysis cannot reflect the complete life cycle of the session is overcome, the problem of semantic gaps caused by limited metadata information dimensions is solved, and the accuracy and robustness of intention recognition are improved.

The invention provides a global optimization method and device for a refinery enterprise supply chain, and the method comprises the steps: obtaining a supplier set, a sales company set, a refinery plant set, a transportation set, a product set and a spatial position set of a refinery enterprise, and constructing a supply chain model, represented by a network graph, of the refinery enterprise; simplifying the supply chain model represented by the network graph by using a graph sampling algorithm to obtain a sampled supply chain model, constructing an effective lower bound function of a supply chain optimization problem, and solving a first optimal solution according to the effective lower bound function and the sampled supply chain model; reducing the supply chain model represented by the network graph by using a graph reduction algorithm to obtain a reduced supply chain model, constructing an effective upper bound function of a supply chain optimization problem, and solving a second optimal solution according to the effective upper bound function and the reduced supply chain model; and generating a global optimal solution of the supply chain model according to the first optimal solution and the second optimal solution.

The application discloses a kind of based on blockchain's federated learning method, device and blockchain node, it is related to blockchain technology field, to solve the problem of poor model precision that existing based on blockchain's federated learning framework is trained.The method comprises: training initial model using own sample until target sample residual that satisfies preset condition is obtained, different alliance chain members have different attribute characteristics;Collaborative training request is issued to blockchain, so that the participants in the alliance chain that respond to collaborative training request collaboratively train blockchain model, and target sample residual is carried in collaborative training request;Collaboratively train blockchain model using common sample with participant, and submit the model parameter obtained after training to blockchain.In this way, the requester publishes the residual of preliminary training in blockchain, so that each participant collaborates to train blockchain model using its unique data characteristics, and higher precision model training can be achieved.

The application discloses a product analysis method and system based on a life cycle-system level data chain model, realizes transparent recording and tracing of product life cycle data through a decentralized data storage platform of a block chain, applies smart contract technology, accurately cooperates global and local decisions, and breaks the collaboration bottleneck of a traditional manufacturing system. Meanwhile, the problem of small sample data imbalance is solved by using a distributed sharing mechanism of the block chain, and the data credibility and the overall availability of the system are improved. On the basis of the decentralized data storage platform, a product life cycle and system level double digital main line block chain data model is established, an intelligent manufacturing data analysis and knowledge learning model is established in a method domain, the block chain is combined with edge computing, data processing is more efficient and intelligent, the intelligent level of the manufacturing system is further improved, and a comprehensive, efficient and intelligent solution is provided for digital transformation.

The invention provides a task execution method and device, equipment, a storage medium and a program product, and the method comprises the steps that a thinking chain model automatically generates instruction information according to task description information and current state information of at least one mode; the instruction execution model obtains instruction information and state information, executes a corresponding instruction and outputs execution information; wherein the state information is updated after the instruction is executed; the matching model determines whether the execution result is matched with the instruction information at least according to the instruction information and the execution information; if the task description information is matched with the task description information, and the updated state information does not conform to the task completion state corresponding to the task description information, the updated state information is used as current state information to be input into the thinking chain model, and the thinking chain model continues to automatically generate next piece of instruction information according to the task description information and the state information; and the state information accords with the task completion state corresponding to the task description information.

The invention provides a logistics industry green innovation performance spatio-temporal heterogeneity analysis method, which comprises the following steps of: 1, constructing a multi-dimensional index system, and calculating a logistics industry green innovation performance comprehensive index through a combined weighting method and a TOPSIS method; 2, coupling the weight matrixes of the geography, economy and logistics networks, and constructing a spatial weight matrix; step 3, analyzing a space driving mechanism of logistics industry green innovation performance by using an improved space Durbin model (SDM), and decomposing space effects into four classes of instant / accumulation and direct / indirect; and 4, constructing a space Markov chain model by taking a space lag factor as a condition, and realizing dynamic prediction and club convergence detection of a logistics industry green innovation performance state. According to the method, thoughts and schemes are provided for solving the problem that static evaluation and dynamic prediction are separated in traditional research, an evaluation-mechanism-prediction framework is formed, and accurate support is provided for formulating regional logistics policies.

The invention discloses a progressive chain interactive document generation method, and belongs to the technical field of thinking chain models. The method mainly comprises the following steps of: 1, analyzing blocks from a given document template; 2, analyzing a core task item, a supplementary description item and an example; 3, requiring the user to provide further related information; 4, taking the question and answer process of the previous step as background data to participate in question and answer of the current step; 5, splicing in the step 2, the step 3 and the step 4 to form a cue word, and generating a thinking process and a reply by a large model; step 6, performing post-processing on a return result of the large model so as to meet a consistency requirement and a structure requirement; 7, modifying a result generated by the model by a user, and confirming that the generated result is available; and step 8, writing the confirmed result in each step into a corresponding position in the template, and outputting a document. According to the progressive chain interactive document generation method, the effect of outputting the thinking process is achieved.

The invention relates to a method for optimizing dimension chain models of sub-parts of a product assembly. The method comprises the steps that the dimension chain models of all the sub-parts are constructed based on three-dimensional geometric data, assembly process parameters and measurement dimension data of the sub-parts; determining simulation size data of the size chain model of the product assembly obtained by assembling the sub-parts based on the size chain model of the sub-parts; measuring actual size data of a product assembly obtained by assembling the sub-parts; and correcting model parameters of the dimension chain model of the sub-part based on first deviation data between the simulated dimension data and the actual dimension data of the product assembly. According to the method and the device, the difference between the simulation size data and the actual size data of the product assembly is reduced in an accelerated manner in a pilot production stage or a product development and debugging stage, the actual size data of the product assembly can meet the dimensional tolerance of the size chain model after less times of tests, the dimensional accuracy of the product assembly is effectively improved, and the product assembly is more accurate. And the problem of size out-of-tolerance in the sub-part assembling process is reduced.

The application provides a kind of multi-task allocation algorithm optimization method for complex task scene, comprising: based on task chain and OODA theory, considering node attribute, heterogeneity and edge directionality on the basis of task chain model, establish effective task net model;According to the effective task net model, a dynamic reconstruction strategy after random failure and deliberate attack failure is established;Initialize the effective task net model, analyze the failure of unmanned cluster nodes and edges, use adjacency matrix and arrival matrix to realize the task success rate of different task allocation algorithms;Through the numerical simulation results of the task success rate of different task allocation algorithms, the advantages and disadvantages of the allocation algorithm are evaluated, and the optimization analysis of the task allocation algorithm under different task scenes is realized.The application greatly improves the task success rate of the cluster in the complex task environment, the task success rate based on the effective task net is simulated and optimized for the task allocation and execution process, effectively supports the control decision of the cluster, and improves the task execution efficiency of the cluster.

The invention provides an outer wall component tolerance distribution optimization method based on multi-tolerance collaboration. The method comprises the following steps: establishing a wallboard two-dimensional analysis system; establishing a dimension chain formula of the wallboard two-dimensional analysis system; an optimization function with the purpose of minimizing the sum of the comprehensive cost coefficients is constructed through correlation coefficients between the tolerances; determining constraint conditions corresponding to the tolerance in the horizontal direction and the vertical direction and constraint conditions of the dimension chain model; an optimal tolerance allocation scheme meeting constraint conditions is obtained through iteration in combination with an improved genetic algorithm containing an elitist strategy; the optimal outer wall component manufacturing tolerance is obtained from the optimal tolerance distribution scheme; according to the relation between the optimal outer wall component manufacturing tolerance and the wallboard two-dimensional analysis system, the sizes of a longitudinal seam, a transverse seam and a diagonal line are obtained; according to the method, multi-tolerance collaboration can be considered, high-accuracy parameters are obtained, and the local optimization problem in the optimization process is avoided by combining the use of an elite strategy.

The invention relates to the technical field of financial risk control and big data, in particular to an enterprise credit dynamic evaluation method and system based on big data, and the method comprises the steps: obtaining commercial operation data, constructing a dynamic transaction graph, and generating a transaction health propagation index; analyzing the real-time operation data flow to obtain a digital asset pledge score; identifying an abnormal transaction group through a cross-chain model; and finally, integrating the indexes to output a dynamic credit evaluation result. According to the method, comprehensive quantitative evaluation is carried out from three dimensions of external ecological risks, internal operation vitality and an associated fraud network, and the defects that a traditional evaluation method is static and isolated, and gang fraud is difficult to identify are overcome.

The invention discloses a cross-chain data sending method and receiving method, electronic equipment and a storage medium. The cross-chain data sending method comprises the following steps: performing block coding on first floating point data to be transmitted to obtain first block floating point data; performing structured packaging on the first block of floating point data to obtain a second block of floating point data; and performing cross-chain transmission on the second block of floating point data to send the second block of floating point data to the second block chain. In a cross-chain scene, data is compressed through block coding, the on-chain storage overhead and the network transmission load of the block chain are reduced, and the data transmission speed and the cross-chain communication efficiency are improved; through structured packaging, the problem of inconsistent analysis of traditional floating point data in a multi-chain environment is avoided, and the exchange efficiency and consistency of the floating point data between multi-chain systems are improved. The method is suitable for cross-chain floating point calculation or data synchronization scenes, in particular to traceability chains, cross-chain model calculation, credible floating point data verification and other scenes with low data precision requirements but large data volume.

To provide a method, device, and a program, capable of accurately predicting stress-strain characteristics of a rubber material in a shorter time and at lower cost, a recording medium, and a design method.SOLUTION: A method for predicting stress-strain characteristics of a rubber material based on an 8-chain model by computer simulation comprises steps of: calculating parameters on the basis of a correlation expression of a molecular chain number n and a segment number N being the parameters of the 8-chain model and at least one of the crosslink density and component formulation amount, created using at least one of the previously-experimentally obtained crosslink density and component formulation amount for a plurality of rubber material samples having different formulations, and previously-experimentally obtained stress-strain characteristics of the rubber material samples, and at least one of the crosslink density and component formulation amount of the rubber material to be analyzed; and predicting the stress-strain characteristics of the rubber material to be analyzed on the basis of the calculated value of the parameters and the 8-chain model.SELECTED DRAWING: Figure 4

Examples related to kinematic joint and chain physical tools for augmented reality implementations are provided. One aspect includes a computing device for kinematic joint simulation, the computing device comprising a display and a processor coupled to a storage system storing instructions that, when executed by the processor, cause the processor to: present a chain simulation interface comprising a plurality of graphical control elements, each graphical control element is configured to adjust at least one physical parameter, receive a selection from a user using a plurality of graphical control elements, update a chain model based on the received selection, and display and animation the updated chain model on a display using a physical engine during kinematic motion simulation, wherein the chain model comprises a plurality of rigid mesh models linked together by spring elements arranged in a tree configuration, thereby forming an assembly skeleton.

The application discloses a Gastrodia elata full-process traceability method and system based on digital twinning and blockchains, and particularly relates to the field of agricultural product quality traceability data processing technology, and is used for solving the problem that the existing traceability system uses a static quality snapshot as the whole-process quality basis, thereby causing the traceability conclusion to deviate from the real quality of the physical object; the method is achieved by the following steps: acquiring initial quality data and environmental parameter sequences after a processing node, simulating active ingredient accumulation and attenuation along the environmental parameter sequences by a digital twinning model to generate a dynamic quality trajectory, separating an irreversible damage component when the environmental parameter sequences abnormally fluctuate, correcting the attenuation factor, and uploading the model state parameters to a blockchain for storage and evidence, when traceability is queried, the dynamic quality trajectory is advanced to the current time point and is consistent with the on-chain model state parameters, physical feasibility is verified based on the ratio of the entropy trap crossing accumulation and the theoretical minimum demand, and according to the verification result, an untrustworthy identifier or a quality change traceability certificate is classified and output.

The application relates to the technical field of encrypted communication and discloses a method and system for intelligent analysis based on multi-source data, wherein the method for intelligent analysis based on multi-source data comprises the following steps: obtaining multi-dimensional metadata of an encrypted communication flow to generate a characteristic vector set; using a Markov chain model to calculate state transition probability of a communication session, dividing the communication session according to a time window and mapping the communication session to discrete states to generate a session state evolution sequence; analyzing the session state evolution sequence based on a hierarchical topic model, mapping the state sequence to a behavior topic through a latent Dirichlet allocation, mapping the behavior topic to an intention category through a hierarchical Dirichlet process, and outputting a communication intention category and a confidence score; the application overcomes the limitation that static feature analysis cannot reflect the complete life cycle of a session, solves the semantic gap problem caused by limited metadata information dimensions, and improves the accuracy and robustness of intention recognition.

The present application relates to the field of material analysis, in particular to a kind of quantitative analysis method of bridging chain mechanics contribution in polymer nanocomposite material.The quantitative analysis method provided by the present application realizes the quantitative analysis of bridging chain mechanics contribution by artificially constructing the model of bridging chain structure, in combination with the accurate tracking of known particle number and local stress decomposition technology, establishes the quantitative relationship between "bridging chain structure-mechanics contribution", provides analysis tool and theoretical basis for rational design of high-performance nanocomposite material.

The application discloses a kind of execution chain segmentation type construction methods for heterogeneous computing platform, it is related to heterogeneous computing and real-time task management technical field.The present application takes the end-to-end execution flow of complete complex application as object, by identifying the basic execution unit in original application execution flow, the execution resource type and effective interaction boundary of each basic execution unit are combined, and the execution chain model composed of multiple execution segments is constructed, and the execution chain model obtained can be directly served for subsequent scheduling analysis, execution control, timing analysis and cross-platform deployment.The present application can accurately describe the real execution process of complex intelligent application on heterogeneous platform, significantly improve the expression ability of execution chain model to end-to-end application flow, improve the integrity and accuracy of heterogeneous execution process modeling, with strong engineering practicability and industrial application prospect.

The invention discloses an auditable chain model digital watermarking method. The method comprises the following steps: an owner inputs a salt value and an owner identifier to a computer system; dividing the model training period into training fragments; calculating the identifier of the current training fragment through a hash function according to the model weight of the previous training fragment; generating current training fragment watermark information by taking the identifier as a seed of a generator; constructing a total loss function containing task loss and watermark loss; watermark loss is realized by constraining a target embedding parameter symbol or a numerical approximation binary watermark bit stream; a watermark is positioned through statistical analysis, and privacy noise is introduced before training fragment storage; storing the model as a detection point file after noise injection as an initial weight of a next fragment; repeating iteration to complete training, and outputting a model hash chain; and according to the verification request, performing dual verification based on the model hash chain, and sending a verification result to a verifier. The method improves the robustness of the watermark to various attacks, and is suitable for model distribution and copyright evidence obtaining scenes.

The invention relates to the field of computer simulation and polymer physics, particularly discloses a simulation method for joint coupling motion of a flexible boundary and an internal active chain based on an annular polymer model, and aims to solve the problem that complex dynamic coupling and nonlinear deformation behaviors between the active chain and the flexible boundary are difficult to really simulate in the prior art. The method comprises the following steps: constructing an annular polymer model formed by connecting inactive particles through chemical bonds so as to characterize a flexible boundary; initializing an active chain model based on boundary geometric constraint, and dynamically aligning driving force and chain segment orientation; adjusting the chain parameters according to the conformation features to generate conformation evolution data; kinetic parameters are extracted to establish a diffusion gradient and volume fraction feedback mechanism, and a bidirectional coupling model is formed by combining the acting force of a local density and velocity field calculation chain on the boundary. According to the technical scheme, non-linear phenomena such as asymmetric collapse, oscillation and bud growth of the boundary can be reproduced in a high-fidelity mode, and the physical authenticity and predictive ability of simulation are remarkably improved.

The application discloses a health improvement method and system based on user behavior patterns, and relates to the technical fields of health management and intelligent device data processing.The method comprises the following steps: S1, continuously collecting multi-dimensional behavior data of a user from at least two different types of intelligent devices, wherein the multi-dimensional behavior data comprises physical activity data, sleep structure data and nutrition intake data, and performing noise removal, normalization and timestamp alignment operations on the multi-dimensional behavior data to generate a standardized behavior data sequence with a minute as the minimum time granularity.The application collects multi-dimensional behavior data from at least two types of intelligent devices, generates a high-quality data sequence through sliding window smoothing, outlier rejection and standardization processing, mines behavior time sequence correlation through dynamic time warping, analyzes causal relationships through improved Granger testing, and constructs a personalized behavior chain model, thereby solving the problem that the prior art can only statistically analyze data in isolation.

The invention discloses a deep learning-based wine appearance authenticity identification method, and particularly relates to the technical field of wine anti-counterfeiting. The method comprises the following steps: collecting multi-angle images of wine products, extracting local disturbance response and global diffusion features, and constructing a pseudo feature response chain model; performing chain-level matching on the chain model and a preset genuine product standard chain library, calculating a similarity score, and judging the authenticity of the product according to a set threshold value; if it is judged that the product is a counterfeit product, reversely reasoning a disguised source area based on chain interruption nodes, and outputting a visual traceability graph; according to the method, a structure disturbance propagation chain modeling mechanism is introduced into wine appearance identification for the first time, and the identification precision and the positioning capability of highly-simulated wine can be remarkably improved.

The application discloses a heterogeneous data cross-chain interaction communication method based on a relay node, the mechanism proposes a cross-chain model architecture composed of a relay node, a relay node chain and a cross-chain mechanism, and designs an IBE-based relay node security access and communication scheme, and the mechanism can be used for realizing safe interaction of data between multiple heterogeneous mechanism chains in an intelligent transportation scene, and providing reliable technical support for data sharing and cooperation in the intelligent transportation. The scheme realizes data communication and cross-chain contract interoperation between different mechanism chains in the intelligent transportation network, improves cross-chain data sharing efficiency, and guarantees the safety of data in the cross-chain process.

The application discloses a kind of multi-party complex protocol chain mode analysis and modeling method based on model driving, the method includes the following steps: step S1, collection, summary and induction protocol in the chain case of upper chain mode;Step S2, upper chain mode extraction;Step S3, build the upper chain model of multi-party service protocol in line with model driving architecture;Step S4, the realization of visual tool.The method can analyze and extract upper chain elements to the blockchain deployment requirement of complex protocol text when system runs, then model the protocol according to upper chain elements, the model meets the standard of model driving architecture, finally realizes the automatic conversion between models without semantic loss, so that the developer of smart contract can better understand business requirements and reduce communication barriers.

The embodiment of the invention discloses a model training method and device based on thinking mode conversion and electronic equipment, and relates to the technical field of natural language processing and machine learning, and the method comprises the steps: taking the completion degree of a target task as a feedback signal, and training an initial thinking chain model through a reinforcement learning and verifiable feedback framework RLVR; sampling a plurality of questions according to the target task, generating thinking processes and answers corresponding to the questions by using a thinking chain model, obtaining a plurality of triads, performing similarity evaluation and sorting, and deleting the thinking processes in the triads of which the similarity meets a set condition to obtain a recombination sequence; training the thinking chain model by using the recombination sequence to obtain a trained thinking chain model; the thinking chain model can be used for autonomously reasoning and automatically converting from slow thinking to fast thinking. The problems that in the prior art, conversion and evolution from slow thinking to fast thinking cannot be automatically achieved, so that task processing efficiency is low, cost is high, and switching flexibility is not enough are solved.

The application discloses a kind of dynamic link access control method and device applied to power system distribution network, the method includes: according to the trusted root data of the platform trusted root and the component data of all platform components, with platform trusted root as trust main line starting point, in combination with the direction of trusted chain control and the expansion rule of trusted chain, obtain the first level trusted chain model, the first level trusted chain model is used to execute trusted verification operation to each platform component;According to component data, determine the corresponding trusted level of each platform component, and according to the trusted level of each platform component, the first level trusted chain model is combined to determine the data control flow, obtain the second level trusted chain model, as the target trusted chain model of data trusted verification operation and data transmission research and judge operation to access data.It can be seen that the application can dynamically adjust the access control according to the real-time permission state of the access terminal, and improve the control security of network access control.

The invention relates to an online book borrowing management method and system based on data analysis, and the method comprises the steps: obtaining the multi-dimensional borrowing value characteristics of all books in a borrowing platform in a preset time period, carrying out the borrowing value evaluation and classification of the books through a borrowing value evaluation model, and obtaining a plurality of borrowing value class groups; a preset user-book borrowing adaptation mechanism is utilized to generate an adaptation value of each book and the to-be-borrowed user, and an initial book recommendation list is obtained and comprises a sub recommendation list corresponding to each borrowing value category; and inputting the initial book recommendation list into a pre-constructed borrowing chain model to obtain target books which generate a borrowing chain effect with the books in the initial book recommendation list, supplementing the target books into the initial book recommendation list to obtain a target recommendation list, and sending the target recommendation list to a borrowing platform of the user to borrow. Therefore, the recommendation accuracy of the book borrowing platform is improved, and the book borrowing success rate and the user stickiness are enhanced.

The application discloses an auditable chain model digital watermarking method, comprising the following steps: an owner inputs a salt value and an owner identifier into a computer system; a model training period is divided into training fragments; a model weight of a previous training fragment is used to calculate an identifier of a current training fragment through a hash function; the identifier is used as a seed of a generator to generate watermark information of the current training fragment; a total loss function containing a task loss and a watermark loss is constructed; watermark loss is realized by constraining a target embedded parameter symbol or a value to approximate a binary watermark bit stream; watermark positioning is realized through statistical analysis; a private noise is introduced before a training fragment is saved; a model saved after noise injection is saved as a detection point file, which is used as initial weight of a next fragment; training is repeated and iterated to output a model hash chain; double verification is performed based on the model hash chain according to a verification request, and a verification result is sent to a verifier. The method improves the robustness of the watermark to various attacks and is suitable for model distribution and copyright evidence scenarios.