Partial-Label Learning with a Reject Option

Tobias Fuchs, Florian Kalinke, Klemens Böhm

In real-world applications, one often encounters ambiguously labeled data, where different annotators assign conflicting class labels. Partial-label learning allows training classi…