Visual Analytics Approaches for Explainable Artificial Intelligence - Detailseite

In recent years machine learning (ML) systems have made significant progress in improving the predictive accuracy for specific tasks and often outperform human users in tasks such as recognizing faces in images. ML systems will become ubiquitous in many areas of human activity because they can make routine work more efficient by relieving humans from time-consuming and recurring parts of their tasks.  We also see the increasing tendency to utilize ML systems to inform high-stakes decisions. In many scenarios, there is a rising awareness that users need to understand how ML systems make particular decisions. Explainable machine learning is a promising research direction to address ML models' often inscrutable black-box nature. In this seminar, we will discuss Visual Analytics approaches (=systems that combine interactive visualization methods and automated analysis methods) that support users to understand the decisions of ML models.