A Case Study on Concept Induction for Neuron-Level Interpretability in CNN

Moumita Sen Sarma, Samatha Ereshi Akkamahadevi, Pascal Hitzler

Deep Neural Networks (DNNs) have advanced applications in domains such as healthcare, autonomous systems, and scene understanding, yet the internal semantics of their hidden neuron…