\textit{FocaLogic}: Logic-Based Interpretation of Visual Model Decisions

TL;DR

FocaLogic is a model-agnostic framework that interprets visual model decisions using logic-based representations, providing transparent, quantitative insights into model behavior and biases.

Contribution

It introduces a novel logic-based approach to interpret visual models, identifying minimal influential regions and providing quantitative evaluation metrics.

Findings

Uncovers training-induced focus concentration.

Shows focus accuracy improves with generalization.

Detects biases and adversarial influences.

Abstract

Interpretability of modern visual models is crucial, particularly in high-stakes applications. However, existing interpretability methods typically suffer from either reliance on white-box model access or insufficient quantitative rigor. To address these limitations, we introduce FocaLogic, a novel model-agnostic framework designed to interpret and quantify visual model decision-making through logic-based representations. FocaLogic identifies minimal interpretable subsets of visual regions-termed visual focuses-that decisively influence model predictions. It translates these visual focuses into precise and compact logical expressions, enabling transparent and structured interpretations. Additionally, we propose a suite of quantitative metrics, including focus precision, recall, and divergence, to objectively evaluate model behavior across diverse scenarios. Empirical analyses demonstrate FocaLogic's capability to uncover critical insights such as training-induced concentration, increasing focus accuracy through generalization, and anomalous focuses under biases and adversarial attacks. Overall, FocaLogic provides a systematic, scalable, and quantitative solution for interpreting visual models.