Learning Dynamic Representations via An Optimally-Weighted Maximum Mean Discrepancy Optimization Framework for Continual Learning

TL;DR

This paper introduces a novel continual learning framework called OWMMD that uses weighted maximum mean discrepancy and adaptive regularization to mitigate catastrophic forgetting and improve model retention of learned information.

Contribution

The paper proposes a new framework combining weighted MMD and adaptive regularization to enhance continual learning and reduce forgetting, with an innovative feature importance assessment mechanism.

Findings

Achieves state-of-the-art performance on benchmarks.

Effectively mitigates catastrophic forgetting.

Outperforms several established baselines.

Abstract

Continual learning has emerged as a pivotal area of research, primarily due to its advantageous characteristic that allows models to persistently acquire and retain information. However, catastrophic forgetting can severely impair model performance. In this study, we address network forgetting by introducing a novel framework termed Optimally-Weighted Maximum Mean Discrepancy (OWMMD), which imposes penalties on representation alterations via a Multi-Level Feature Matching Mechanism (MLFMM). Furthermore, we propose an Adaptive Regularization Optimization (ARO) strategy to refine the adaptive weight vectors, which autonomously assess the significance of each feature layer throughout the optimization process, The proposed ARO approach can relieve the over-regularization problem and promote the future task learning. We conduct a comprehensive series of experiments, benchmarking our proposed method against several established baselines. The empirical findings indicate that our approach achieves state-of-the-art performance.