Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GOAL: Geometrically Optimal Alignment for Continual Generalized Category Discovery

Jizhou Han, Chenhao Ding, SongLin Dong, Yuhang He, Shaokun Wang, Qiang Wang, Yihong Gong

TL;DR

GOAL, a unified framework that introduces a fixed Equiangular Tight Frame (ETF) classifier to impose a consistent geometric structure throughout learning, is proposed, enabling stable integration of new classes without disrupting old ones.

Abstract

Continual Generalized Category Discovery (C-GCD) requires identifying novel classes from unlabeled data while retaining knowledge of known classes over time. Existing methods typically update classifier weights dynamically, resulting in forgetting and inconsistent feature alignment. We propose GOAL, a unified framework that introduces a fixed Equiangular Tight Frame (ETF) classifier to impose a consistent geometric structure throughout learning. GOAL conducts supervised alignment for labeled samples and confidence-guided alignment for novel samples, enabling stable integration of new classes without disrupting old ones. Experiments on four benchmarks show that GOAL outperforms the prior method Happy, reducing forgetting by 16.1% and boosting novel class discovery by 3.2%, establishing a strong solution for long-horizon continual discovery.

GOAL: Geometrically Optimal Alignment for Continual Generalized Category Discovery

TL;DR

GOAL, a unified framework that introduces a fixed Equiangular Tight Frame (ETF) classifier to impose a consistent geometric structure throughout learning, is proposed, enabling stable integration of new classes without disrupting old ones.

Abstract

Continual Generalized Category Discovery (C-GCD) requires identifying novel classes from unlabeled data while retaining knowledge of known classes over time. Existing methods typically update classifier weights dynamically, resulting in forgetting and inconsistent feature alignment. We propose GOAL, a unified framework that introduces a fixed Equiangular Tight Frame (ETF) classifier to impose a consistent geometric structure throughout learning. GOAL conducts supervised alignment for labeled samples and confidence-guided alignment for novel samples, enabling stable integration of new classes without disrupting old ones. Experiments on four benchmarks show that GOAL outperforms the prior method Happy, reducing forgetting by 16.1% and boosting novel class discovery by 3.2%, establishing a strong solution for long-horizon continual discovery.
Paper Structure (24 sections, 13 equations, 3 figures, 4 tables)

This paper contains 24 sections, 13 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Generalized Category Discovery
  4. Continual Generalized Category Discovery
  5. Neural Collapse and Geometric Structures
  6. Preliminaries
  7. Problem Setting of C-GCD
  8. Definition of Optimal Geometric Structure
  9. Method
  10. Overview
  11. Base Session
  12. Predefined ETF Prototypes
  13. Supervised Alignment to ETF Prototypes
  14. Representation Learning and Parametric Classification
  15. Incremental Sessions
  16. ...and 9 more sections

Figures (3)

  • Figure 1: Conceptual illustration of the Continual Generalized Category Discovery (C-GCD) task and classifier design strategies. (a) Overview of the C-GCD task: a model is trained on labeled base classes and incrementally discovers novel categories from unlabeled data in subsequent sessions. (b) Prior studies typically use learnable prototypes or classifier weights. (c) Our method adopts a fixed optimal geometric structure, ensuring better alignment and generalization across continual sessions.
  • Figure 2: Overview of the GOAL framework. In the base session, labeled features are aligned to fixed ETF prototypes via supervised training. In incremental sessions, high-confidence unlabeled samples are aligned to unallocated prototypes with unsupervised learning. This maintains geometric consistency for continual discovery and retention.
  • Figure 3: Accuracy trend across different values of $\alpha$ on CIFAR100.