Professor Shan Suthaharan has had a research paper titled “Comparison of Explainable AI Techniques for Early Detection of Potential Prediction Failures” accepted at the IEEE 27th International Conference on Information Reuse and Integration for Data Science (IRI 2026), a long-standing and competitive international IEEE conference in artificial intelligence and data science.
The paper introduces and evaluates nEGXAI-V, a vision-based extension of the original negation-based explainable artificial intelligence (XAI) framework, now referred to as nEGXAI-T for tabular data. Building on the negation-based foundation of the original framework, the proposed methods focus on analyzing both the strongest positive confidence and the competing negative-class confidence in convolutional neural network (CNN) models.
The research further demonstrates how nEGXAI-V delivers additional knowledge that enhances human interpretability through positive evidence, negative evidence, contrastive importance, and nEGXAI-fused explanations, contributing toward more reliable, transparent, and trustworthy AI systems.
The IEEE 27th International Conference on Information Reuse and Integration for Data Science (IEEE IRI 2026) will be held in Seattle, Washington, from July 31 to August 2, 2026.
Congratulations to Professor Shan Suthaharan!
