Faster Verified Explanations for Neural Networks
By: Alessandro De Palma, Greta Dolcetti, Caterina Urban
Potential Business Impact:
Makes AI's decisions understandable and trustworthy.
Verified explanations are a theoretically-principled way to explain the decisions taken by neural networks, which are otherwise black-box in nature. However, these techniques face significant scalability challenges, as they require multiple calls to neural network verifiers, each of them with an exponential worst-case complexity. We present FaVeX, a novel algorithm to compute verified explanations. FaVeX accelerates the computation by dynamically combining batch and sequential processing of input features, and by reusing information from previous queries, both when proving invariances with respect to certain input features, and when searching for feature assignments altering the prediction. Furthermore, we present a novel and hierarchical definition of verified explanations, termed verifier-optimal robust explanations, that explicitly factors the incompleteness of network verifiers within the explanation. Our comprehensive experimental evaluation demonstrates the superior scalability of both FaVeX, and of verifier-optimal robust explanations, which together can produce meaningful formal explanation on networks with hundreds of thousands of non-linear activations.
Similar Papers
Explaining, Fast and Slow: Abstraction and Refinement of Provable Explanations
Machine Learning (CS)
Makes AI predictions understandable and trustworthy.
Faithful Counterfactual Visual Explanations (FCVE)
CV and Pattern Recognition
Shows how computers "see" to make them smarter.
Calibrated Reasoning: An Explanatory Verifier for Dynamic and Efficient Problem-Solving
Artificial Intelligence
Helps computers check their own answers better.