InfoSculpt: Sculpting the Latent Space for Generalized Category Discovery
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908767401017344 |
|---|---|
| author | Liao, Wenwen Ruan, Hang Yu, Jianbo Wang, Yuansong Jiang, Qingchao Yang, Xiaofeng |
| author_facet | Liao, Wenwen Ruan, Hang Yu, Jianbo Wang, Yuansong Jiang, Qingchao Yang, Xiaofeng |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_10098 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | InfoSculpt: Sculpting the Latent Space for Generalized Category Discovery Liao, Wenwen Ruan, Hang Yu, Jianbo Wang, Yuansong Jiang, Qingchao Yang, Xiaofeng Computer Vision and Pattern Recognition 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. |
| title | InfoSculpt: Sculpting the Latent Space for Generalized Category Discovery |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2601.10098 |