HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts

TL;DR

This paper introduces HiLo, a novel learning framework for generalized category discovery that is robust to domain shifts, effectively handling unlabelled data from multiple domains and outperforming existing models.

Contribution

The paper proposes HiLo, a new method that extracts semantic and domain features separately, incorporates domain augmentation and curriculum learning, and establishes a benchmark for GCD under domain shifts.

Findings

HiLo outperforms state-of-the-art models on all evaluations.

The method effectively handles multi-domain unlabelled data.

Benchmark results demonstrate robustness to real-world domain shifts.

Abstract

Generalized Category Discovery (GCD) is a challenging task in which, given a partially labelled dataset, models must categorize all unlabelled instances, regardless of whether they come from labelled categories or from new ones. In this paper, we challenge a remaining assumption in this task: that all images share the same domain. Specifically, we introduce a new task and method to handle GCD when the unlabelled data also contains images from different domains to the labelled set. Our proposed `HiLo' networks extract High-level semantic and Low-level domain features, before minimizing the mutual information between the representations. Our intuition is that the clusterings based on domain information and semantic information should be independent. We further extend our method with a specialized domain augmentation tailored for the GCD task, as well as a curriculum learning approach. Finally, we construct a benchmark from corrupted fine-grained datasets as well as a large-scale evaluation on DomainNet with real-world domain shifts, reimplementing a number of GCD baselines in this setting. We demonstrate that HiLo outperforms SoTA category discovery models by a large margin on all evaluations.