Dissecting Generalized Category Discovery: Multiplex Consensus under Self-Deconstruction
This addresses the challenge of enabling machines to recognize objects in both known and novel categories, which is incremental as it builds on existing GCD methods with a new paradigm.
The paper tackles the problem of generalized category discovery by proposing ConGCD, a method that decomposes objects into visual primitives and uses multiplex consensus to integrate diverse cues, achieving state-of-the-art results on benchmarks.
Human perceptual systems excel at inducing and recognizing objects across both known and novel categories, a capability far beyond current machine learning frameworks. While generalized category discovery (GCD) aims to bridge this gap, existing methods predominantly focus on optimizing objective functions. We present an orthogonal solution, inspired by the human cognitive process for novel object understanding: decomposing objects into visual primitives and establishing cross-knowledge comparisons. We propose ConGCD, which establishes primitive-oriented representations through high-level semantic reconstruction, binding intra-class shared attributes via deconstruction. Mirroring human preference diversity in visual processing, where distinct individuals leverage dominant or contextual cues, we implement dominant and contextual consensus units to capture class-discriminative patterns and inherent distributional invariants, respectively. A consensus scheduler dynamically optimizes activation pathways, with final predictions emerging through multiplex consensus integration. Extensive evaluations across coarse- and fine-grained benchmarks demonstrate ConGCD's effectiveness as a consensus-aware paradigm. Code is available at github.com/lytang63/ConGCD.