Lab data security protection system fusing federated learning and reinforcement learning

By using gradient orthogonal decomposition and dynamic network structure adjustment of edge intelligent agents, combined with reputation assessment and global gradient feedback of federated collaborative servers, the problems of heterogeneous laboratory data adaptability and privacy leakage in existing technologies are solved, thereby improving the robustness and resource efficiency of laboratory data security protection systems.

CN121664522BActive Publication Date: 2026-06-16BEIJING NORMAL UNIVERSITY
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING NORMAL UNIVERSITY
Filing Date
2025-12-10
Publication Date
2026-06-16

AI Technical Summary

⚠Technical Problem

Existing distributed security protection technologies cannot adapt to the characteristics of heterogeneous laboratory data. Gradient transmission is prone to leaking the privacy of original data, and the model adaptability is poor due to the scarcity of samples and malicious nodes.

⚗Method used

By using edge intelligent agents to perform gradient orthogonal decomposition and dynamic network structure adjustment locally, combined with the reputation assessment of federated collaborative servers and global gradient residual feedback, the data security protection system achieves adaptability and privacy protection.

🎯Benefits of technology

It effectively cuts off the gradient backpropagation path, improves the model's robustness and generalization ability in identifying new types of attacks, reduces computational resource consumption, and ensures the security and stability of laboratory data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121664522B_ABST
    Figure CN121664522B_ABST
Patent Text Reader

Abstract

The application relates to the technical field of network security and distributed artificial intelligence, and discloses a laboratory data security protection system fusing federated learning and reinforcement learning, which comprises an edge intelligent agent, a federated collaborative server and an encrypted communication network. The edge intelligent agent utilizes a manifold topological resonance module to dynamically adjust a strategy network structure according to data complexity, generates virtual samples for auxiliary training through an adversarial course enhancement module, and separates local private gradients and shared gradients by utilizing a gradient orthogonal decomposition module. The federated collaborative server is responsible for gradient weighted aggregation and global attention parameter distribution based on node reputation. The application realizes high-privacy collaborative training in a distributed environment by adapting heterogeneous data features through geometric manifolds, cutting off gradient privacy association through orthogonal projection and making up for sample scarcity through adversarial generation, helps to solve the problems of poor model adaptability and privacy leakage in laboratory data protection, and improves the security and adaptability of the system.
Need to check novelty before this filing date? Find Prior Art