TARGET: Federated Class-Continual Learning via Exemplar-Free Distillation
This addresses the challenge of catastrophic forgetting in federated learning with non-IID data, which is incremental as it builds on existing FCCL methods by removing the need for external data or data storage.
The paper tackles the problem of Federated Class-Continual Learning (FCCL), where new classes are added dynamically in federated learning, by proposing TARGET, a method that alleviates catastrophic forgetting without requiring additional datasets or storing private data, achieving improved performance in data-sensitive scenarios.
This paper focuses on an under-explored yet important problem: Federated Class-Continual Learning (FCCL), where new classes are dynamically added in federated learning. Existing FCCL works suffer from various limitations, such as requiring additional datasets or storing the private data from previous tasks. In response, we first demonstrate that non-IID data exacerbates catastrophic forgetting issue in FL. Then we propose a novel method called TARGET (federat\textbf{T}ed cl\textbf{A}ss-continual lea\textbf{R}nin\textbf{G} via \textbf{E}xemplar-free dis\textbf{T}illation), which alleviates catastrophic forgetting in FCCL while preserving client data privacy. Our proposed method leverages the previously trained global model to transfer knowledge of old tasks to the current task at the model level. Moreover, a generator is trained to produce synthetic data to simulate the global distribution of data on each client at the data level. Compared to previous FCCL methods, TARGET does not require any additional datasets or storing real data from previous tasks, which makes it ideal for data-sensitive scenarios.