Composing Novel Classes: A Concept-Driven Approach to Generalized Category Discovery

2024-10-17

arXiv.org

AI Summary

To address insufficient utilization of known-class knowledge in Generalized Category Discovery (GCD), this paper proposes ConceptGCD—a concept-driven two-stage learning framework. Methodologically, it leverages a pretrained model on known classes to extract discriminative semantic concepts, then transfers knowledge to novel classes via generative concept expansion, contrastive constraints, and covariance regularization. Key contributions include: (1) the first differentiable/non-differentiable concept dichotomy mechanism, enabling explicit semantic concept modeling with gradient-controllable optimization; and (2) a covariance-enhanced generative loss coupled with concept-score normalization, facilitating hierarchical concept disentanglement and balanced learning. Extensive experiments on multiple benchmarks demonstrate significant improvements over state-of-the-art methods, validating the critical role of concept-level knowledge transfer in unsupervised novel category discovery. (138 words)

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We tackle the generalized category discovery (GCD) problem, which aims to discover novel classes in unlabeled datasets by leveraging the knowledge of known classes. Previous works utilize the known class knowledge through shared representation spaces. Despite their progress, our analysis experiments show that novel classes can achieve impressive clustering results on the feature space of a known class pre-trained model, suggesting that existing methods may not fully utilize known class knowledge. To address it, we introduce a novel concept learning framework for GCD, named ConceptGCD, that categorizes concepts into two types: derivable and underivable from known class concepts, and adopts a stage-wise learning strategy to learn them separately. Specifically, our framework first extracts known class concepts by a known class pre-trained model and then produces derivable concepts from them by a generator layer with a covariance-augmented loss. Subsequently, we expand the generator layer to learn underivable concepts in a balanced manner ensured by a concept score normalization strategy and integrate a contrastive loss to preserve previously learned concepts. Extensive experiments on various benchmark datasets demonstrate the superiority of our approach over the previous state-of-the-art methods. Code will be available soon.

Problem

• Discover novel classes in unlabeled datasets
• Utilize known class knowledge effectively
• Introduce ConceptGCD for improved category discovery

Innovation

• ConceptGCD framework for novel class discovery
• Stage-wise learning of derivable and underivable concepts
• Covariance-augmented loss and concept score normalization