Interpreting deep learning output for out-of-distribution detection

Damian Matuszewski, Ida-Maria Sintorn

Commonly used AI networks are very self-confident in their predictions, even when the evidence for a certain decision is dubious. The investigation of a deep learning model output…