Continual Learning for non-stationary regression via Memory-Efficient Replay
This addresses the need for efficient continual learning in dynamic environments like Industry 4.0, focusing on regression tasks where research is limited, though it appears incremental as it adapts existing replay methods to a new task type.
The paper tackles the problem of continual learning for regression tasks in non-stationary data streams, proposing a prototype-based generative replay framework that reduces forgetting and achieves more stable performance than state-of-the-art methods on benchmark datasets.
Data streams are rarely static in dynamic environments like Industry 4.0. Instead, they constantly change, making traditional offline models outdated unless they can quickly adjust to the new data. This need can be adequately addressed by continual learning (CL), which allows systems to gradually acquire knowledge without incurring the prohibitive costs of retraining them from scratch. Most research on continual learning focuses on classification problems, while very few studies address regression tasks. We propose the first prototype-based generative replay framework designed for online task-free continual regression. Our approach defines an adaptive output-space discretization model, enabling prototype-based generative replay for continual regression without storing raw data. Evidence obtained from several benchmark datasets shows that our framework reduces forgetting and provides more stable performance than other state-of-the-art solutions.