文献类型：期刊文献

英文题名：InfoSculpt: Sculpting the Latent Space for Generalized Category Discovery

作者：Liao, Wenwen[1]; Ruan, Hang[1]; Yu, Jianbo[2]; Wang, Yuansong[3]; Jiang, Qingchao[4]; Yang, Xiaofeng[2]

机构：[1] College of Intelligent Robotics and Advance Manufacturing, Fudan University, China; [2] School of Microelectronics, Fudan University, China; [3] Tsinghua Shenzhen International Graduate School, Tsinghua University, China; [4] School of Information Science and Engineering, East China University of Science and Technology, China

年份：2026

外文期刊名：arXiv

收录：EI(收录号：20260058512)

语种：英文

外文关键词：Information theory - Large datasets

摘要：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. Code will be released after acceptance. Copyright ? 2026, The Authors. All rights reserved.