Project

Fail-Controlled Classifiers

Safety-aware classifiers that can reject uncertain predictions for dependable AI systems.

Designed multiple fail-controlled classifier architectures, evaluated them on image and tabular datasets, and published results in dependable AI venues.