Unsupervised Learning of Neural Networks to Explain Neural Networks

By Quanshi Zhang, Yu Yang, Ying Nian Wu, Song-Chun Zhu. 2018

29 citations recorded by the source corpus.

This paper presents an unsupervised method to learn a neural network, namely an explainer, to interpret a pre-trained convolutional neural network (CNN), i.e., explaining knowledge representations hidden in middle conv-layers of the CNN. Given feature maps of a certain conv-layer of the CNN, the explainer performs like an auto-encoder, which first disentangles the feature maps into object-part features and then inverts object-part features back to features of higher conv-layers of the CNN. More specifically, the explainer contains interpretable conv-layers, where each filter disentangles the representation of a specific object part from chaotic input feature maps. As a paraphrase of CNN features, the disentangled representations of object parts help people understand the logic inside the CNN. We also learn the explainer to use object-part features to reconstruct features of higher CNN layers, in order to minimize loss of information during the feature disentanglement. More crucially, we learn the explainer via network distillation without using any annotations of sample labels, object parts, or textures for supervision. We have applied our method to different types of CNNs for evaluation, and explainers have significantly boosted the interpretability of CNN features.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Song-Chun Zhu
2. Quanshi Zhang
3. Ying Nian Wu
4. Zhi Chen
5. Yu Yang
6. Cynthia Rudin
7. Leilani H. Gilpin
8. Yijie Bei

In Computer Vision and Pattern Recognition

1. Song-Chun Zhu
2. Quanshi Zhang
3. Ying Nian Wu
4. Yu Yang
5. Cynthia Rudin
6. Zhi Chen
7. Alan L. Yuille
8. Luc Van Gool

In Safety, Robustness & Interpretability

1. Quanshi Zhang
2. Song-Chun Zhu
3. Zhi Chen
4. Ying Nian Wu
5. Cynthia Rudin
6. Yu Yang
7. Yijie Bei
8. Leilani H. Gilpin

Citations in the corpus

• Explaining Explanations: An Overview of Interpretability of Machine Learning (2018, background): “Disentangled units can enable the construction of graphs [68] and decision trees [69] to elucidate the reasoning of a network.”
• Interpreting CNNs via Decision Trees (2018, methodology): “[41] used an explainer network to interpret object-part representations in intermediate layers of CNNs.”
• Concept whitening for interpretable image recognition (2020, methodology): “Zhang et al. (2018) add losses to the filters to encourage them to detect object parts that can also be viewed as concepts.”
• Explaining Explanations: An Approach to Evaluating Interpretability of Machine Learning (2018, background): “Disentangled units can enable the construction of graphs [68] and decision trees [69] to elucidate the reasoning of a network.”