Explanatory Paradigms in Neural Networks: Towards relevant and contextual explanations
Ghassan AlRegib, Mohit Prabhushankar · IEEE Signal Processing Magazine · 2022
In this article, we present a leap-forward expansion to the study of explainability in neural networks by considering explanations as answers to abstract reasoning-based questions. With P as the prediction from a neural network, these questions are “WhyP?”, “What if notP?”, and “WhyP, rather thanQ?” for a given contrast predictionQ. The answers to these questions are observed correlations, counterfactuals, and contrastive explanations, respectively. Together, these explanations constitute the abductive reasoning scheme. The term observed refers to the specific case of posthoc explainability when an explanatory technique explains the decisionPafter a trained neural network has made the decision. The primary advantage of viewing explanations through the lens of abductive reasoning-based questions is that explanations can be used as reasons while making decisions. The posthoc field of explainability, which previously justified decisions, becomes active by being involved in the decision-making process and providing limited but relevant and contextual interventions. The contributions of this article are 1) realizing explanations as reasoning paradigms, 2) providing a probabilistic definition of observed explanations and their completeness, 3) creating a taxonomy for evaluation of explanations, and 4) positioning gradient-based complete explainability’s replicability and reproducibility across multiple applications and data modalities, and 5) code repositories, which are publicly available athttps://github.com/olivesgatech/Explanatory-Paradigms.