Space Explanations of Neural Network Classification
This addresses the need for interpretable AI by offering a novel explanation method with provable guarantees, though it appears incremental as an improvement over existing logic-based approaches.
The paper tackles the problem of explaining neural network classifications by introducing Space Explanations, a logic-based concept that provides provable guarantees of network behavior across continuous input feature areas. The authors demonstrate through real-life case studies that their automatically generated explanations are more meaningful than state-of-the-art methods.
We present a novel logic-based concept called Space Explanations for classifying neural networks that gives provable guarantees of the behavior of the network in continuous areas of the input feature space. To automatically generate space explanations, we leverage a range of flexible Craig interpolation algorithms and unsatisfiable core generation. Based on real-life case studies, ranging from small to medium to large size, we demonstrate that the generated explanations are more meaningful than those computed by state-of-the-art.