InfoSculpt: Sculpting the Latent Space for Generalized Category Discovery

TL;DR

This paper introduces InfoSculpt, an information-theoretic framework that improves generalized category discovery by disentangling category signals from noise in the latent space, leading to more robust and accurate classification.

Contribution

We propose InfoSculpt, a novel method based on the Information Bottleneck principle that systematically sculpts the latent space for better category discovery in large-scale unlabeled data.

Findings

Outperforms existing methods on 8 benchmark datasets.

Effectively disentangles category information from instance noise.

Produces a more robust and discriminative representation space.

Abstract

Generalized Category Discovery (GCD) aims to classify instances from both known and novel categories within a large-scale unlabeled dataset, a critical yet challenging task for real-world, open-world applications. However, existing methods often rely on pseudo-labeling, or two-stage clustering, which lack a principled mechanism to explicitly disentangle essential, category-defining signals from instance-specific noise. In this paper, we address this fundamental limitation by re-framing GCD from an information-theoretic perspective, grounded in the Information Bottleneck (IB) principle. We introduce InfoSculpt, a novel framework that systematically sculpts the representation space by minimizing a dual Conditional Mutual Information (CMI) objective. InfoSculpt uniquely combines a Category-Level CMI on labeled data to learn compact and discriminative representations for known classes, and a complementary Instance-Level CMI on all data to distill invariant features by compressing augmentation-induced noise. These two objectives work synergistically at different scales to produce a disentangled and robust latent space where categorical information is preserved while noisy, instance-specific details are discarded. Extensive experiments on 8 benchmarks demonstrate that InfoSculpt validating the effectiveness of our information-theoretic approach.