Rehearsal-free Federated Domain-incremental Learning

2024-05-22

arXiv.org

AI Summary

This paper addresses catastrophic forgetting in domain-incremental federated learning. We propose a rehearsal-free, domain-incremental federated learning framework that requires no data replay and stores no local historical data. Our method introduces two key innovations: (1) a novel global prompt-sharing mechanism enabling cross-client knowledge collaboration; and (2) a domain-adaptive prompt generator coupled with a domain-specific prompt contrastive loss, jointly optimizing domain-invariant feature learning and fine-grained local prompt modeling to enhance cross-domain knowledge retention and discriminability. The approach incurs zero additional memory overhead—critical for resource-constrained and privacy-sensitive settings—while significantly improving domain identification accuracy and model stability. Extensive experiments demonstrate superior performance over existing federated continual learning methods in terms of both generalization across domains and robustness to forgetting.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Prompt Engineering / Prompting
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

We introduce a rehearsal-free federated domain incremental learning framework, RefFiL, based on a global prompt-sharing paradigm to alleviate catastrophic forgetting challenges in federated domain-incremental learning, where unseen domains are continually learned. Typical methods for mitigating forgetting, such as the use of additional datasets and the retention of private data from earlier tasks, are not viable in federated learning (FL) due to devices' limited resources. Our method, RefFiL, addresses this by learning domain-invariant knowledge and incorporating various domain-specific prompts from the domains represented by different FL participants. A key feature of RefFiL is the generation of local fine-grained prompts by our domain adaptive prompt generator, which effectively learns from local domain knowledge while maintaining distinctive boundaries on a global scale. We also introduce a domain-specific prompt contrastive learning loss that differentiates between locally generated prompts and those from other domains, enhancing RefFiL's precision and effectiveness. Compared to existing methods, RefFiL significantly alleviates catastrophic forgetting without requiring extra memory space, making it ideal for privacy-sensitive and resource-constrained devices.

Problem

• Addresses catastrophic forgetting in federated domain-incremental learning
• Eliminates need for rehearsal data in resource-constrained FL settings
• Enhances domain adaptation via global prompt-sharing and contrastive learning

Innovation

• Rehearsal-free federated domain-incremental learning framework
• Global prompt-sharing paradigm for domain-invariant knowledge
• Domain-specific prompt contrastive learning loss