Rehearsal-Free Continual Learning over Small Non-I.I.D. Batches
This addresses the challenge of enabling robots to learn continuously from short, non-i.i.d. video batches without forgetting, which is crucial for real-world adaptation but incremental in method.
The paper tackles the problem of continual learning for robotic vision by introducing a new protocol based on the CORe50 benchmark and proposing two rehearsal-free techniques, CWR* and AR1*, which achieve over 15% higher accuracy than other state-of-the-art methods in some cases when trained on nearly 400 small non-i.i.d. incremental batches.
Robotic vision is a field where continual learning can play a significant role. An embodied agent operating in a complex environment subject to frequent and unpredictable changes is required to learn and adapt continuously. In the context of object recognition, for example, a robot should be able to learn (without forgetting) objects of never before seen classes as well as improving its recognition capabilities as new instances of already known classes are discovered. Ideally, continual learning should be triggered by the availability of short videos of single objects and performed on-line on on-board hardware with fine-grained updates. In this paper, we introduce a novel continual learning protocol based on the CORe50 benchmark and propose two rehearsal-free continual learning techniques, CWR* and AR1*, that can learn effectively even in the challenging case of nearly 400 small non-i.i.d. incremental batches. In particular, our experiments show that AR1* can outperform other state-of-the-art rehearsal-free techniques by more than 15% accuracy in some cases, with a very light and constant computational and memory overhead across training batches.