Can we Constrain Concept Bottleneck Models to Learn Semantically Meaningful Input Features?

Re-assign community