From Segments to Concepts: Interpretable Image Classification via Concept-Guided Segmentation

2025-10-05

AI Summary

Deep neural networks face dual challenges of interpretability and robustness in safety-critical vision tasks due to their black-box decision-making and reliance on spurious correlations. To address this, we propose SEG-MIL-CBM—a novel concept-bottleneck framework that integrates concept-guided image segmentation into attention-driven multi-instance learning (MIL), enabling spatially anchored concept reasoning without concept-level annotations. Through end-to-end optimization, the model automatically localizes image regions supporting high-level semantic concepts and establishes pixel-level, interpretable alignments between regions and concepts. Unlike conventional concept bottleneck models (CBMs), SEG-MIL-CBM eliminates dependence on manual concept annotations and global average pooling, thereby significantly improving robustness against spurious correlations and input noise. Extensive experiments on multiple benchmarks demonstrate that SEG-MIL-CBM achieves superior classification accuracy and concept-level interpretability compared to state-of-the-art explainable methods.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Deep neural networks have achieved remarkable success in computer vision; however, their black-box nature in decision-making limits interpretability and trust, particularly in safety-critical applications. Interpretability is crucial in domains where errors have severe consequences. Existing models not only lack transparency but also risk exploiting unreliable or misleading features, which undermines both robustness and the validity of their explanations. Concept Bottleneck Models (CBMs) aim to improve transparency by reasoning through human-interpretable concepts. Still, they require costly concept annotations and lack spatial grounding, often failing to identify which regions support each concept. We propose SEG-MIL-CBM, a novel framework that integrates concept-guided image segmentation into an attention-based multiple instance learning (MIL) framework, where each segmented region is treated as an instance and the model learns to aggregate evidence across them. By reasoning over semantically meaningful regions aligned with high-level concepts, our model highlights task-relevant evidence, down-weights irrelevant cues, and produces spatially grounded, concept-level explanations without requiring annotations of concepts or groups. SEG-MIL-CBM achieves robust performance across settings involving spurious correlations (unintended dependencies between background and label), input corruptions (perturbations that degrade visual quality), and large-scale benchmarks, while providing transparent, concept-level explanations.

Problem

• Addresses lack of interpretability in deep neural network image classification
• Identifies relevant image regions without requiring concept annotations
• Mitigates spurious correlations and input corruptions for robust performance

Innovation

• Concept-guided segmentation for interpretable image classification
• Attention-based multiple instance learning without concept annotations
• Spatially grounded explanations via segmented region aggregation