Intrinsic Concept Extraction Based on Compositional Interpretability
This work addresses the inability to extract composable object-level and attribute-level concepts from single images for unsupervised concept extraction, which is incremental as it builds on diffusion-based models and hierarchical modeling.
The paper tackled the problem of extracting composable intrinsic concepts from a single image, which existing methods fail to do, by introducing the CI-ICE task and proposing HyperExpress, a method that achieves accurate concept disentanglement and maintains complex inter-concept relationships, demonstrating outstanding performance.
Unsupervised Concept Extraction aims to extract concepts from a single image; however, existing methods suffer from the inability to extract composable intrinsic concepts. To address this, this paper introduces a new task called Compositional and Interpretable Intrinsic Concept Extraction (CI-ICE). The CI-ICE task aims to leverage diffusion-based text-to-image models to extract composable object-level and attribute-level concepts from a single image, such that the original concept can be reconstructed through the combination of these concepts. To achieve this goal, we propose a method called HyperExpress, which addresses the CI-ICE task through two core aspects. Specifically, first, we propose a concept learning approach that leverages the inherent hierarchical modeling capability of hyperbolic space to achieve accurate concept disentanglement while preserving the hierarchical structure and relational dependencies among concepts; second, we introduce a concept-wise optimization method that maps the concept embedding space to maintain complex inter-concept relationships while ensuring concept composability. Our method demonstrates outstanding performance in extracting compositionally interpretable intrinsic concepts from a single image.