Explanatory Interactive Machine Learning

Abstract The most promising standard machine learning methods can deliver highly accurate classification results, often outperforming white-box methods. However, it is hardly possible for humans to fully understand the rationale behind black-box and thus, these powerful hamper creation of new knowledge on part broader acceptance this technology. Explainable Artificial Intelligence attempts overcome problem by making results more interpretable, while Interactive Machine Learning integrates into process insight discovery. paper builds recent successes in combining two cutting-edge technologies proposes how Explanatory (XIL) embedded a generalizable Action Design Research (ADR) – called XIL-ADR. This approach be used analyze data, inspect models, iteratively improve them. shows application using diagnosis viral pneumonia, e.g., Covid-19, as an illustrative example. By means, also illustrates XIL-ADR help identify shortcomings projects, gain insights human user, thereby unlock full potential AI-based systems organizations research.