Code Division Modulation Layers Against Forgetting and Inference in Continual Gait Identification

2026-07-21

AI Summary

This work addresses the challenges of catastrophic forgetting and vulnerability to membership inference attacks in continual learning for gait recognition, which are exacerbated by reliance on data replay and its associated communication overhead. To overcome these limitations, the study introduces, for the first time, a code-division modulation layer to construct a replay-free continual learning framework. This approach effectively mitigates catastrophic forgetting and enhances privacy protection without requiring retransmission of historical data. Experimental results demonstrate that the proposed method maintains high recognition accuracy across sequential tasks while significantly reducing the success rate of membership inference attacks, thereby achieving a favorable balance between model performance and security.

Abstract

Continual learning (CL) has been recently employed in biometric identification systems thanks to its ability to integrate new knowledge within a pre-trained model and to the possibility of reducing the computational cost of training. Unfortunately, such approaches pose new challenges both in terms of final accuracy and privacy guarantees since a progressive fine-tuning of the model on small subsets expose them to catastrophic forgetting and successful inference attacks. This paper evaluates the efficiency of code division modulation layers (CDML) on a gait identification system which has been trained following a continual learning policy. The proposed approach preserves accuracy on all the tasks while mitigating membership inference attacks at the same time. Moreover, the impact of retransmission is minimized since replaying data is not necessary.

Problem

• continual learning
• catastrophic forgetting
• membership inference attacks
• gait identification
• privacy

Innovation

• Code Division Modulation Layers
• Continual Learning
• Gait Identification
• Catastrophic Forgetting
• Membership Inference Attacks