CUAL: Continual Uncertainty-aware Active Learner
This addresses the challenge of enabling AI systems to adapt continuously to novelties post-deployment in domains like autonomous systems or robotics, representing an incremental advance in combining continual learning with active learning for novelty detection.
The paper tackles the problem of continual adaptation for AI agents in real-world settings where unlabeled data may contain both known and novel classes, with only a small labeling budget available. The result is the CUAL model, which uses uncertainty estimation to prioritize labeling ambiguous novel class samples and pseudo-labeling certain predictions, showing effectiveness across multiple datasets and backbones.
AI deployed in many real-world use cases should be capable of adapting to novelties encountered after deployment. Here, we consider a challenging, under-explored and realistic continual adaptation problem: a deployed AI agent is continuously provided with unlabeled data that may contain not only unseen samples of known classes but also samples from novel (unknown) classes. In such a challenging setting, it has only a tiny labeling budget to query the most informative samples to help it continuously learn. We present a comprehensive solution to this complex problem with our model "CUAL" (Continual Uncertainty-aware Active Learner). CUAL leverages an uncertainty estimation algorithm to prioritize active labeling of ambiguous (uncertain) predicted novel class samples while also simultaneously pseudo-labeling the most certain predictions of each class. Evaluations across multiple datasets, ablations, settings and backbones (e.g. ViT foundation model) demonstrate our method's effectiveness. We will release our code upon acceptance.