A novel interaction-based methodology towards explainable AI with better understanding of Pneumonia Chest X-ray Images

Abstract

Abstract In the field of eXplainable AI (XAI), robust “blackbox” algorithms such as Convolutional Neural Networks (CNNs) are known for making high prediction performance. However, the ability to explain and interpret these algorithms still require innovation in the understanding of influential and, more importantly, explainable features that directly or indirectly impact the performance of predictivity. A number of methods existing in literature focus on visualization techniques but the concepts of explainability and interpretability still require rigorous definition. In view of the above needs, this paper proposes an interaction-based methodology–Influence score (I-score)—to screen out the noisy and non-informative variables in the images hence it nourishes an environment with explainable and interpretable features that are directly associated to feature predictivity. The selected features with high I-score values can be considered as a group of variables with interactive effect, hence the proposed name interaction-based methodology. We apply the proposed method on a real world application in Pneumonia Chest X-ray Image data set and produced state-of-the-art results. We demonstrate how to apply the proposed approach for more general big data problems by improving the explainability and interpretability without sacrificing the prediction performance. The contribution of this paper opens a novel angle that moves the community closer to the future pipelines of XAI problems. In investigation of Pneumonia Chest X-ray Image data, the proposed method achieves 99.7% Area-Under-Curve (AUC) using less than 20,000 parameters while its peers such as VGG16 and its upgraded versions require at least millions of parameters to achieve on-par performance. Using I-score selected explainable features allows reduction of over 98% of parameters while delivering same or even better prediction results.