A Self-explaining Neural Architecture for Generalizable Concept Learning

arXiv:2405.00349v1 Announce Type: new
Abstract: With the wide proliferation of Deep Neural Networks in high-stake applications, there is a growing demand for explainability behind their decision-making process. Concept learning models attempt to learn high-level 'concepts' - abstract entities that align with human understanding, and thus provide interpretability to DNN architectures. However, in this paper, we demonstrate that present SOTA concept learning approaches suffer from two major problems - lack of concept fidelity wherein the models fail to learn consistent concepts …

abstract applications architecture architectures arxiv concept concepts cs.lg decision demand dnn explainability however human interpretability learn making networks neural networks process type understanding