From Model Patterns to Abstract Semantics in Compositional Zero-Shot Learning

arXiv:2609.15649v1 Announce Type: new Abstract: Compositional Zero Shot Learning aims to recognize unseen compositions by recombining learned primitives. Recent methods rely on vision language models and attempt to explicitly model contextual variations of primitives through multiple representations. However, such approaches are limited by fixed variant capacity and competition between abstract and concrete semantics. In this work, we present a new perspective that views primitive variations as the context-driven activation of concrete visual cues rather than independent entities. Based on it, we propose CLEAR, a CLoze-style rEAsoning-based Re-ranking framework inspired by human perceptual processes. CLEAR extracts conditional variants from the primitive candidate set in a coarse-to-fine manner, performs cloze-style reasoning to infer high-level semantics, and re-ranks predictions to correct biases toward salient concrete primitives. Extensive experiments demonstrate that CLEAR consistently improves the Base Model and outperforms state-of-the-art methods on the challenging C-GQA and MIT-States datasets. Code is available at https://github.com/buptLwz/CLEAR.

Related

• Progressive Reasoning with Primitive Correction for Compositional Zero-Shot Learning
• Attributes Should Come from Images, Not Class Names: Distribution-Conditioned Attribute Selection for Vision-Language Models
• Sparse Autoencoders enable Robust and Interpretable Fine-tuning of CLIP models

Source: arXiv cs.CV | 2026-09-15