Search

CN-121997379-A - AI model production optimization method and system based on blockchain and meta-universe data markers

CN121997379ACN 121997379 ACN121997379 ACN 121997379ACN-121997379-A

Abstract

The application relates to the technical field of artificial intelligence and blockchain, and discloses an AI model production optimization system based on blockchain and metauniverse data marking, which comprises a blockchain network module, a zero trust architecture module and a zero trust architecture module, wherein the blockchain network module is used for storing hash values of marking data generated in a metauniverse environment and related metadata in a distributed manner, realizing automatic verification and access authority control of the marking data through intelligent contracts, and integrating the zero trust architecture module into a blockchain network, and is used for verifying the validity of an access request in real time and adjusting the access authority of a user according to a dynamic authority management strategy in a blockchain. The distributed storage and the consensus verification of the meta-universe data are realized through the block chain network module, the backup and the consistency of the data on a plurality of nodes are ensured, the non-falsification of the block chain ensures the integrity and the authenticity of all recorded data and operation logs, and the risk of falsifying the data is effectively prevented, so that the safety of data management is improved.

Inventors

  • LIU XIAOWEN

Assignees

  • 刘晓雯

Dates

Publication Date
20260508
Application Date
20241101

Claims (10)

  1. 1. AI model production optimization system based on blockchain and metauniverse data mark, characterized by including: the block chain network module is used for storing hash values of the marking data generated in the meta-universe environment and related metadata in a distributed mode, and realizing automatic verification and access right control of the marking data through intelligent contracts; the zero trust architecture module is integrated in the blockchain network, and is used for verifying the validity of the access request in real time and adjusting the access authority of the user according to a dynamic authority management strategy in the blockchain; The AI model training module is used for carrying out dynamic feature extraction and self-adaptive model training on the verified data, and managing traceability and automation of parameter optimization in the model training process based on the marking data and the metadata of the blockchain records; And the intelligent contract module is used for managing the version control, the automatic deployment and the monitoring of the model performance of the AI model, and recording and verifying the updating process of the model through the blockchain network.
  2. 2. The AI model production optimization system based on blockchain and metauniverse data markers of claim 1 wherein the blockchain network module further includes a rights management unit for managing associations between public keys of nodes and their operating rights and for automating verification of data writing and reading operations using smart contracts implemented by: receiving an access request, and extracting a public key in the request; searching corresponding rights according to the extracted public key; and if the authority allows, performing data writing or reading operation, and recording the operation result.
  3. 3. The AI model production optimization system based on blockchain and metauniverse data markers of claim 1 wherein the zero trust architecture module further comprises a dynamic rights adjustment unit for monitoring and analyzing user behavior, the dynamic rights adjustment unit automatically adjusting its access rights based on the user's history of operation in the blockchain network, the adjustment process comprising the steps of: Recording hash values and related metadata of each operation of a user; Analyzing the frequency, type and result of user operation; And generating a right adjustment strategy according to the analysis result, and executing the strategy through the intelligent contract.
  4. 4. The AI model production optimization system based on blockchain and metauniverse data markers of claim 1 wherein the smart contract module further includes an automatic rollback unit that automatically performs the following steps when it is detected that model performance does not meet preset criteria: retrieving model parameters of a previous version and hash values thereof; replacing the current version of the model with the retrieved model parameters; updating the model version record on the blockchain to ensure the transparency and consistency of the rollback operation.
  5. 5. The AI model production optimization method based on the blockchain and the metauniverse data mark is characterized by comprising the following steps of: generating original data in a meta-universe environment, marking the data by using an algorithm, generating marked data and calculating a hash value of the marked data; step two, data storage, namely storing the hash value of the marking data and metadata thereof in a blockchain network; Verifying the legitimacy of the access request through a zero trust architecture, and adjusting the access authority of the user according to a dynamic authority management strategy in the blockchain; Step four, feature extraction and model training, namely carrying out dynamic feature extraction and self-adaptive model training on the marked data, and carrying out model training and parameter optimization based on the data recorded by the blockchain; And fifthly, model version control and deployment, namely carrying out hash processing and storage on parameters of the AI model by utilizing the intelligent contract, and managing version control and automatic deployment of the model.
  6. 6. The blockchain and metauniverse data marker-based AI model production optimization method of claim 5, wherein the AI model training module performs feature extraction using a multi-layer Convolutional Neural Network (CNN), the feature extraction process comprising: A first layer convolution operation is applied to the input data D i , where the convolution formula is: X 1 ＝Conv 1 (D i )＝W 1 ·D i +b 1 And processing a first layer convolution result by applying a ReLU activation function, wherein an activation formula is as follows: X' 1 ＝ReLU(X 1 ) And applying a second-layer convolution operation to the activation result, wherein the convolution formula is as follows: X 2 ＝Conv 1 (X' 1 )＝W 2 ·X' 1 +b 2 And finally outputting a characteristic vector X i for subsequent model training.
  7. 7. The AI model production optimization method based on blockchain and metauniverse data markers of claim 5, wherein the access control in step three comprises the steps of: Receiving an access request of a user and generating a hash value of the request; storing the generated hash value and the requested related metadata in a blockchain; And verifying the access authority of the user through the zero trust architecture module, and allowing or rejecting access according to the verification result.
  8. 8. The AI model production optimization method based on blockchain and metauniverse data markers of claim 5, wherein the model training in step four comprises the following: Model training is performed by using the feature vector X i as input; Applying a back propagation algorithm during training, updating the model parameters according to the following formula: Wherein, eta is the learning rate, Gradient of the loss function with respect to model parameters; and performing adaptive optimization of the model by using the marking data and the metadata of the blockchain record.
  9. 9. The AI model production optimization method based on blockchain and metauniverse data markers of claim 5, wherein model version control in step five comprises the steps of: generating a hash value for each trained model version, and storing the hash value in a blockchain together with metadata such as version numbers, time stamps and the like; automatically triggering the intelligent contract to verify the performance of the model, and recording a verification result in a block chain; and determining whether the model is deployed or rolled back to the previous version according to the verification result.
  10. 10. The AI model production optimization method based on blockchain and metauniverse data markers of claim 5, wherein the data marker algorithm in step one comprises the steps of: automatically marking the original data in the meta-universe environment based on a predefined rule to generate initial marking data; The final marking data is generated by user interaction correction or supplement of the marking data; The hash value of the final tag data is calculated and stored with the associated metadata in the blockchain.

Description

AI model production optimization method and system based on blockchain and meta-universe data markers Technical Field The invention relates to the technical field of artificial intelligence and blockchain, in particular to an AI model production optimization method and system based on blockchain and meta-universe data marking. Background With the rapid development of Artificial Intelligence (AI) technology, more and more industries are beginning to employ AI models for data analysis, prediction, and decision support. The success of AI models relies on a large amount of high quality training data, while the accuracy, integrity and safety of the data directly affects the performance of the model. The meta space (METAVERSE) serves as a virtual and reality fused digital space, and generates a large amount of user interaction data and virtual environment data. These data can provide rich material for training of AI models. However, since the generation environment of meta-universe data is diverse and complex, it is difficult for the conventional centralized data storage and management method to ensure the authenticity and security of data. In addition, potential security holes exist in the processes of marking, storing, accessing and controlling the data, the data is easy to leak or falsify, and the training effect and the application safety of the AI model are further affected. Meanwhile, along with the wide application of the AI model in the key fields of finance, medical treatment, automatic driving and the like, the training and deployment process of the AI model also faces higher requirements of transparency, traceability and automatic management. Conventional model training methods often have difficulty tracking each step and parameter update in the model training process, resulting in an inability to effectively monitor and verify the behavior of the model after it is deployed. In addition, model version control, automatic deployment and performance verification are also important problems in AI model application, and the traditional method has low automation degree in the aspects, relies on manual intervention, and is easy to cause misoperation or delay. Disclosure of Invention Aiming at the defects of the prior art, the invention provides an AI model production optimization method and system based on blockchain and meta universe data markers, which solve the problems of unsafe data management, high risk of data tampering, non traceability of model training process, dependence on manual intervention for model deployment and low degree of automation in the prior art. In order to achieve the above purpose, the invention is realized by the following technical scheme that the AI model production optimization system based on blockchain and metauniverse data marking comprises: the block chain network module is used for storing hash values of the marking data generated in the meta-universe environment and related metadata in a distributed mode, and realizing automatic verification and access right control of the marking data through intelligent contracts; the zero trust architecture module is integrated in the blockchain network, and is used for verifying the validity of the access request in real time and adjusting the access authority of the user according to a dynamic authority management strategy in the blockchain; The AI model training module is used for carrying out dynamic feature extraction and self-adaptive model training on the verified data, and managing traceability and automation of parameter optimization in the model training process based on the marking data and the metadata of the blockchain records; And the intelligent contract module is used for managing the version control, the automatic deployment and the monitoring of the model performance of the AI model, and recording and verifying the updating process of the model through the blockchain network. Preferably, the blockchain network module further includes a rights management unit for managing association between a public key of a node and its operation rights, and performing automated verification of data writing and reading operations using an intelligent contract, the intelligent contract being implemented by: receiving an access request, and extracting a public key in the request; searching corresponding rights according to the extracted public key; and if the authority allows, performing data writing or reading operation, and recording the operation result. Preferably, the zero trust architecture module further comprises a dynamic authority adjustment unit for monitoring and analyzing user behavior, the dynamic authority adjustment unit automatically adjusts the access authority of the user according to the operation history of the user in the blockchain network, and the adjustment process comprises the following steps: Recording hash values and related metadata of each operation of a user; Analyzing the frequency, type and result of u