Maintain Plasticity in Long-timescale Continual Test-time Adaptation

TL;DR

This paper investigates the challenge of maintaining model plasticity during long-term continual test-time adaptation, proposing a novel method to preserve adaptability and improve performance in non-stationary environments.

Contribution

It introduces the Adaptive Shrink-Restore (ASR) policy that re-initializes weights based on label flip changes to sustain plasticity over time.

Findings

Plasticity declines steadily in long-term CTTA.

Loss of plasticity correlates with label flip changes.

ASR method effectively preserves plasticity and improves adaptation performance.

Abstract

Continual test-time domain adaptation (CTTA) aims to adjust pre-trained source models to perform well over time across non-stationary target environments. While previous methods have made considerable efforts to optimize the adaptation process, a crucial question remains: can the model adapt to continually-changing environments with preserved plasticity over a long time? The plasticity refers to the model's capability to adjust predictions in response to non-stationary environments continually. In this work, we explore plasticity, this essential but often overlooked aspect of continual adaptation to facilitate more sustained adaptation in the long run. First, we observe that most CTTA methods experience a steady and consistent decline in plasticity during the long-timescale continual adaptation phase. Moreover, we find that the loss of plasticity is strongly associated with the change in label flip. Based on this correlation, we propose a simple yet effective policy, Adaptive Shrink-Restore (ASR), towards preserving the model's plasticity. In particular, ASR does the weight re-initialization by the adaptive intervals. The adaptive interval is determined based on the change in label flipping. Our method is validated on extensive CTTA benchmarks, achieving excellent performance.