Expert Routing with Synthetic Data for Continual Learning

TL;DR

This paper introduces G2D, a continual learning method that uses synthetic data to train a domain discriminator, enabling effective expert routing across domains without data sharing, outperforming existing methods in vision and language tasks.

Contribution

Proposes G2D, a novel synthetic data-based domain-discriminator for expert routing in continual learning, demonstrating superior performance over existing methods.

Findings

G2D outperforms competitive domain-incremental methods.

Synthetic data is more effective for training domain discriminators than direct classifier training.

Applicable to both vision and language modalities.

Abstract

In many real-world settings, regulations and economic incentives permit the sharing of models but not data across institutional boundaries. In such scenarios, practitioners might hope to adapt models to new domains, without losing performance on previous domains (so-called catastrophic forgetting). While any single model may struggle to achieve this goal, learning an ensemble of domain-specific experts offers the potential to adapt more closely to each individual institution. However, a core challenge in this context is determining which expert to deploy at test time. In this paper, we propose Generate to Discriminate (G2D), a domain-incremental continual learning method that leverages synthetic data to train a domain-discriminator that routes samples at inference time to the appropriate expert. Surprisingly, we find that leveraging synthetic data in this capacity is more effective than using the samples to \textit{directly} train the downstream classifier (the more common approach to leveraging synthetic data in the lifelong learning literature). We observe that G2D outperforms competitive domain-incremental learning methods on tasks in both vision and language modalities, providing a new perspective on the use of synthetic data in the lifelong learning literature.