Explaining Neural Networks Semantically and Quantitatively

TL;DR

This paper introduces a method to interpret CNN predictions both semantically and quantitatively by distilling knowledge into an explainable additive model, enhancing understanding and practical application of neural networks.

Contribution

The paper proposes a novel knowledge distillation approach to create an explainable additive model from CNNs, addressing bias issues and providing quantitative explanations.

Findings

Effective in explaining CNN predictions

Addresses bias-interpreting problems in explainable models

Demonstrates improved interpretability in experiments

Abstract

This paper presents a method to explain the knowledge encoded in a convolutional neural network (CNN) quantitatively and semantically. The analysis of the specific rationale of each prediction made by the CNN presents a key issue of understanding neural networks, but it is also of significant practical values in certain applications. In this study, we propose to distill knowledge from the CNN into an explainable additive model, so that we can use the explainable model to provide a quantitative explanation for the CNN prediction. We analyze the typical bias-interpreting problem of the explainable model and develop prior losses to guide the learning of the explainable additive model. Experimental results have demonstrated the effectiveness of our method.