可解释的组合卷积神经网络

Jul, 2021

Interpretable Compositional Convolutional Neural Networks

Wen Shen, Zhihua Wei, Shikun Huang, Binbin Zhang, Jiaqi Fan...

TL;DR通过修改传统卷积神经网络，将其转化为可解释的组合卷积神经网络，以学习中间卷积层中编码有意义的视觉模式的滤波器，从而实现语义可解释的AI，该方法可以广泛应用于不同类型的CNN，并且实验效果良好。

Abstract

The reasonable definition of semantic interpretability presents the core challenge in explainable AI. This paper proposes a method to modify a traditional convolutional neural network (CNN) into an interpretable composi