Empirical Characterization of Rationale Stability Under Controlled Perturbations for Explainable Pattern Recognition

Researchers propose a new metric to assess consistency of AI model explanations across similar inputs, implementing it on BERT models for sentiment analysis. The framework uses cosine similarity of SHAP values to detect inconsistent reasoning patterns and biased feature reliance, providing more robust evaluation of model behavior.

• →Novel metric developed to quantify consistency of AI model explanations across samples with same class labels.
• →Framework implemented using BERT, RoBERTa, and DistilBERT models on sentiment analysis datasets with SHAP feature importance.
• →Method uses cosine similarity to detect biased reliance on features and inconsistent reasoning patterns.
• →Approach enables more robust verification of rationale stability in explainable AI systems.
• →Research code publicly available for reproducibility and further development.