GlanceNets: Interpretable, Leak-proof Concept-based Models Emanuele Marconato emanuele.marconato@unitn.it 1 2 3 Andrea Passerini andrea.passerini@unitn.it 2 3 Stefano Teso stefano.teso@unitn.it 0 3 CIMeC, University of Trento , Italy DII, University of Pisa , Italy DISI, University of Trento , Italy The 17th International Workshop on Neural-Symbolic Learning and Reasoning

In this extended abstract, we briefly outline GlanceNets [1], a new class of deep learning classifiers that acquire high-level concepts from data and use them for both computing predictions and generating ante-hoc explanations of those predictions. In contrast with other concept-based networks, GlanceNets ensure the learned concepts, and the explanations built on them, are human interpretable, even in out-of-distribution scenarios. The core ideas at the heart of GlanceNets extend naturally to other Neuro-Symbolic architectures involving reasoning during inference.

 eol>Concept-based Models Concept Learning Representation Learning Explainable AI