Interactive Classification 0: Overview

I am writing a series of posts on the subject on formal AI interpretability with focus on interpretability approaches that apply to neural networks leading up to our recent work on Merlin-Arthur Classifiers.

1. Preliminaries
2. Prover-Verifier Setup
3. Asymmetric Feature Correlation
4. Relative Strength between Cooperative and Adversarial Prover
5. Computational Complexity