ReLKD: Inter-Class Relation Learning with Knowledge Distillation for Generalized Category Discovery

TL;DR

ReLKD introduces a novel framework for generalized category discovery that leverages implicit inter-class relations through hierarchical and distillation modules, significantly improving classification of both known and novel classes.

Contribution

The paper proposes ReLKD, an end-to-end method that exploits inter-class relations via hierarchical and distillation modules to enhance GCD performance.

Findings

Effective in scenarios with limited labeled data

Outperforms previous methods on four datasets

Utilizes hierarchical class relations for better generalization

Abstract

Generalized Category Discovery (GCD) faces the challenge of categorizing unlabeled data containing both known and novel classes, given only labels for known classes. Previous studies often treat each class independently, neglecting the inherent inter-class relations. Obtaining such inter-class relations directly presents a significant challenge in real-world scenarios. To address this issue, we propose ReLKD, an end-to-end framework that effectively exploits implicit inter-class relations and leverages this knowledge to enhance the classification of novel classes. ReLKD comprises three key modules: a target-grained module for learning discriminative representations, a coarse-grained module for capturing hierarchical class relations, and a distillation module for transferring knowledge from the coarse-grained module to refine the target-grained module's representation learning. Extensive experiments on four datasets demonstrate the effectiveness of ReLKD, particularly in scenarios with limited labeled data. The code for ReLKD is available at https://github.com/ZhouF-ECNU/ReLKD.