Generalized Category Discovery via Token Manifold Capacity Learning

TL;DR

This paper introduces MTMC, a novel method for generalized category discovery that maximizes token manifold capacity to improve clustering accuracy and semantic richness in open-world scenarios.

Contribution

It proposes the Maximum Token Manifold Capacity (MTMC) approach, emphasizing manifold capacity maximization for better intra-class diversity and discriminability.

Findings

MTMC outperforms existing GCD methods in accuracy.

MTMC improves category number estimation.

MTMC reduces dimensional collapse and enhances representations.

Abstract

Generalized category discovery (GCD) is essential for improving deep learning models' robustness in open-world scenarios by clustering unlabeled data containing both known and novel categories. Traditional GCD methods focus on minimizing intra-cluster variations, often sacrificing manifold capacity, which limits the richness of intra-class representations. In this paper, we propose a novel approach, Maximum Token Manifold Capacity (MTMC), that prioritizes maximizing the manifold capacity of class tokens to preserve the diversity and complexity of data. MTMC leverages the nuclear norm of singular values as a measure of manifold capacity, ensuring that the representation of samples remains informative and well-structured. This method enhances the discriminability of clusters, allowing the model to capture detailed semantic features and avoid the loss of critical information during clustering. Through theoretical analysis and extensive experiments on coarse- and fine-grained datasets, we demonstrate that MTMC outperforms existing GCD methods, improving both clustering accuracy and the estimation of category numbers. The integration of MTMC leads to more complete representations, better inter-class separability, and a reduction in dimensional collapse, establishing MTMC as a vital component for robust open-world learning. Code is in github.com/lytang63/MTMC.