Joshua Lockhart

Learn to explain yourself, when you can: Equipping Concept Bottleneck Models with the ability to abstain on their concept predictions

2022 · arXiv (Cornell University)

The Concept Bottleneck Models (CBMs) of Koh et al. [2020] provide a means to ensure that a neural network based classifier bases its predictions solely on human understandable concepts. The concept labels, or rationales as …