ProPaLL: Probabilistic Partial Label Learning

TL;DR

ProPaLL is a new probabilistic method for partial label learning that simplifies training, enhances performance, and is adaptable to any deep learning architecture, demonstrating superior results on various datasets.

Contribution

It introduces ProPaLL, a probabilistic approach that advances partial label learning by simplifying training and improving accuracy across architectures.

Findings

ProPaLL outperforms existing methods on artificial datasets.

ProPaLL achieves better results on real-world datasets.

The approach is compatible with any deep architecture.

Abstract

Partial label learning is a type of weakly supervised learning, where each training instance corresponds to a set of candidate labels, among which only one is true. In this paper, we introduce ProPaLL, a novel probabilistic approach to this problem, which has at least three advantages compared to the existing approaches: it simplifies the training process, improves performance, and can be applied to any deep architecture. Experiments conducted on artificial and real-world datasets indicate that ProPaLL outperforms the existing approaches.