Lateral channel analysis method and system for moment limitation regularization self-adaptive optimization

By introducing a moment-limited regularization adaptive optimization method in side channel analysis, the problem that existing optimization algorithms are prone to overfitting is solved, and the model is better generalization ability and training stability are achieved.

CN119691594AActive Publication Date: 2025-03-25BEIHANG UNIV
3 Cites 0 Cited by

Patent Information

Application Number
CN202411858395.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-17
Publication Date
2025-03-25
Estimated Expiration
2044-12-17

AI Technical Summary

Technical Problem

Existing optimization algorithms are prone to overfitting problems in side channel analysis, resulting in the model performing well on the training set but poor generalization ability on the test set.

Method used

A moment-limited regularization adaptive optimization method is proposed. By explicitly introducing regularization constraints in the optimization process, it suppresses over-update model parameters, thereby improving the generalization ability of the model. This method seamlessly combines the effect of regularization with the gradient update process by dynamically adjusting the regularization intensity.

Benefits of technology

This method can not only reduce the test performance degradation caused by the optimizer overfitting the training data, but also improve the robustness of noise during training, improve the generalization ability and training stability of the model.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a moment limit regularization self-adaptive optimization side channel analysis method and a moment limit regularization self-adaptive optimization side channel analysis system. The method comprises the following steps: acquiring a plurality of pieces of side channel trace data, and recombining and dividing; preprocessing the divided side channel trace data to obtain processed data; constructing a side channel analysis model, and classifying and identifying the processed data to obtain the probability of a predicted label value of each piece of side channel trace data; predicting the probability of the label value based on the obtained side channel trace data, and optimizing the side channel analysis model to obtain a final model; and completing side channel analysis by using the final model. According to the method, the model training stability and generalization ability are improved through the adaptive learning rate and the regularization term, overfitting is avoided, the convergence speed is increased, and the effect is remarkable especially when data are limited or the model is complex.
Need to check novelty before this filing date? Find Prior Art