Interactive Bias Detection in Supervised Classification Models

We present an interactive tool for detecting and diagnosing biases in supervised classification models. Through coordinated views and drill-down capabilities, users can identify which subgroups are most affected and trace bias sources to training data characteristics.