Continuous Unsupervised Domain Adaptation Using Stabilized Representations and Experience Replay

TL;DR

This paper presents a novel unsupervised domain adaptation algorithm for continual learning that stabilizes internal representations and uses experience replay to prevent forgetting, enabling effective adaptation to new domains with only unlabeled data.

Contribution

The paper introduces a new method combining stabilized internal representations modeled by GMMs with experience replay for continual unsupervised domain adaptation, addressing the challenge of domain shift without labeled data.

Findings

Effective in maintaining model performance across domains

Outperforms existing UDA methods in continual learning scenarios

Reduces catastrophic forgetting through experience replay

Abstract

We introduce an algorithm for tackling the problem of unsupervised domain adaptation (UDA) in continual learning (CL) scenarios. The primary objective is to maintain model generalization under domain shift when new domains arrive continually through updating a base model when only unlabeled data is accessible in subsequent tasks. While there are many existing UDA algorithms, they typically require access to both the source and target domain datasets simultaneously. Conversely, existing CL approaches can handle tasks that all have labeled data. Our solution is based on stabilizing the learned internal distribution to enhances the model generalization on new domains. The internal distribution is modeled by network responses in hidden layer. We model this internal distribution using a Gaussian mixture model (GMM ) and update the model by matching the internally learned distribution of new domains to the estimated GMM. Additionally, we leverage experience replay to overcome the problem of catastrophic forgetting, where the model loses previously acquired knowledge when learning new tasks. We offer theoretical analysis to explain why our algorithm would work. We also offer extensive comparative and analytic experiments to demonstrate that our method is effective. We perform experiments on four benchmark datasets to demonstrate that our approach is effective.