Category Adaptation Meets Projected Distillation in Generalized Continual Category Discovery
This addresses the challenge of continual learning with partial labels for AI systems needing to adapt to new categories without forgetting old ones, representing an incremental improvement over existing methods.
The paper tackles the problem of learning from sequentially arriving, partially labeled datasets in Generalized Continual Category Discovery (GCCD), where traditional feature distillation methods restrict adaptability to new categories. The result is a novel technique called CAMP that integrates a learnable projector with feature distillation and an auxiliary category adaptation network, significantly improving the balance between learning new information and retaining old knowledge, with superior performance across several GCCD and Class Incremental Learning scenarios.
Generalized Continual Category Discovery (GCCD) tackles learning from sequentially arriving, partially labeled datasets while uncovering new categories. Traditional methods depend on feature distillation to prevent forgetting the old knowledge. However, this strategy restricts the model's ability to adapt and effectively distinguish new categories. To address this, we introduce a novel technique integrating a learnable projector with feature distillation, thus enhancing model adaptability without sacrificing past knowledge. The resulting distribution shift of the previously learned categories is mitigated with the auxiliary category adaptation network. We demonstrate that while each component offers modest benefits individually, their combination - dubbed CAMP (Category Adaptation Meets Projected distillation) - significantly improves the balance between learning new information and retaining old. CAMP exhibits superior performance across several GCCD and Class Incremental Learning scenarios. The code is available at https://github.com/grypesc/CAMP.