Large-scale Pre-trained Models are Surprisingly Strong in Incremental Novel Class Discovery
This addresses the challenge of lifelong learning for AI systems by enabling truly unsupervised incremental class discovery, though it is incremental as it builds on existing pre-trained models.
The paper tackles the problem of continuously discovering novel classes in unlabelled datasets without needing any related labelled data, by leveraging self-supervised pre-trained models, and shows that simple baselines with a frozen backbone and linear classifier outperform sophisticated state-of-the-art methods in extensive benchmarks.
Discovering novel concepts in unlabelled datasets and in a continuous manner is an important desideratum of lifelong learners. In the literature such problems have been partially addressed under very restricted settings, where novel classes are learned by jointly accessing a related labelled set (e.g., NCD) or by leveraging only a supervisedly pre-trained model (e.g., class-iNCD). In this work we challenge the status quo in class-iNCD and propose a learning paradigm where class discovery occurs continuously and truly unsupervisedly, without needing any related labelled set. In detail, we propose to exploit the richer priors from strong self-supervised pre-trained models (PTM). To this end, we propose simple baselines, composed of a frozen PTM backbone and a learnable linear classifier, that are not only simple to implement but also resilient under longer learning scenarios. We conduct extensive empirical evaluation on a multitude of benchmarks and show the effectiveness of our proposed baselines when compared with sophisticated state-of-the-art methods. The code is open source.