Understanding, Visualizing and Explaining XAI Through Case Studies

Abstract

Data-driven AI applications require Accountability, Responsibility, and Transparency. At the same time,users' expectations on the accuracy and precision of the predictions are unreasonable. This draws attention to - explainability, an inherent limitation in the traditional intelligent systems that are black box in nature. As Explainable AI(XAI) seeks to bridge this gap through the explanation of AI models, this paper focuses on the study of achieving Responsible AI through XAI. AIX360 is a robust popular tool that is used to provide explainability from various viewpoints. Hence, this work uses AIX 360 to bring explainability in the viewpoint of various stakeholders namely domain experts, service providers and customers. The uniqueness of the paper lies in designing a conceptual model for XAI based predictive analysis. This paper is an attempt to understand, visualize and explain the observations through two data-intensive intelligence systems as case studies HELOC(Home Equity Line Of Credit) and ISIC(International Skin Imaging Collaboration).

Ähnliche Arbeiten