Parametric Information Maximization for Generalized Category Discovery

TL;DR

This paper presents a Parametric Information Maximization model for Generalized Category Discovery, effectively handling class imbalance and achieving state-of-the-art results across diverse datasets, including fine-grained categories.

Contribution

The paper introduces a bi-level optimization framework that explores a family of objective functions to improve GCD performance, especially on imbalanced datasets.

Findings

Consistently outperforms existing methods on six datasets.

Effectively handles both short-tailed and long-tailed data.

Achieves new state-of-the-art results in fine-grained GCD tasks.

Abstract

We introduce a Parametric Information Maximization (PIM) model for the Generalized Category Discovery (GCD) problem. Specifically, we propose a bi-level optimization formulation, which explores a parameterized family of objective functions, each evaluating a weighted mutual information between the features and the latent labels, subject to supervision constraints from the labeled samples. Our formulation mitigates the class-balance bias encoded in standard information maximization approaches, thereby handling effectively both short-tailed and long-tailed data sets. We report extensive experiments and comparisons demonstrating that our PIM model consistently sets new state-of-the-art performances in GCD across six different datasets, more so when dealing with challenging fine-grained problems.