CSTA: Spatial-Temporal Causal Adaptive Learning for Exemplar-Free Video Class-Incremental Learning

TL;DR

This paper introduces a causal adaptive learning framework for exemplar-free video class-incremental learning, effectively preserving spatial-temporal knowledge and reducing conflicts to achieve state-of-the-art results.

Contribution

It proposes a novel causal perspective with distillation and compensation modules to improve exemplar-free continual learning for videos.

Findings

Achieves 4.2% higher accuracy than existing methods on benchmarks.

Effectively maintains spatial-temporal knowledge during incremental learning.

Reduces conflicts between different information types through causal mechanisms.

Abstract

Continual learning aims to acquire new knowledge while retaining past information. Class-incremental learning (CIL) presents a challenging scenario where classes are introduced sequentially. For video data, the task becomes more complex than image data because it requires learning and preserving both spatial appearance and temporal action involvement. To address this challenge, we propose a novel exemplar-free framework that equips separate spatiotemporal adapters to learn new class patterns, accommodating the incremental information representation requirements unique to each class. While separate adapters are proven to mitigate forgetting and fit unique requirements, naively applying them hinders the intrinsic connection between spatial and temporal information increments, affecting the efficiency of representing newly learned class information. Motivated by this, we introduce two key innovations from a causal perspective. First, a causal distillation module is devised to maintain the relation between spatial-temporal knowledge for a more efficient representation. Second, a causal compensation mechanism is proposed to reduce the conflicts during increment and memorization between different types of information. Extensive experiments conducted on benchmark datasets demonstrate that our framework can achieve new state-of-the-art results, surpassing current example-based methods by 4.2% in accuracy on average.