POEM: Pattern-Oriented Explanations of Convolutional Neural Networks-Reference-Cited by-同舟云学术

POEM: Pattern-Oriented Explanations of Convolutional Neural Networks

Published:2023-07 Issue:11 Volume:16 Page:3192-3200
ISSN:2150-8097
Container-title:Proceedings of the VLDB Endowment
language:en
Short-container-title:Proc. VLDB Endow.

Author:

Dadvar Vargha¹,Golab Lukasz¹,Srivastava Divesh²

Affiliation:

1. University of Waterloo

2. AT&T Chief Data Office

Abstract

Convolutional Neural Networks (CNNs) are commonly used in computer vision. However, their predictions are difficult to explain, as is the case with many deep learning models. To address this problem, we present POEM, a modular framework that produces patterns of semantic concepts such as shapes and colours to explain image classifier CNNs. POEM identifies patterns such as "if sofa then living room", meaning that if an image contains a sofa and the model pays attention to the sofa, then the model classifies the image as a living room. We illustrate the advantages of POEM over existing work using quantitative and qualitative experiments.

Publisher

Association for Computing Machinery (ACM)

Subject

General Earth and Planetary Sciences,Water Science and Technology,Geography, Planning and Development

Link

https://dl.acm.org/doi/pdf/10.14778/3611479.3611518

Reference30 articles.

1. Arjun R. Akula , Keze Wang , Changsong Liu , Sari Saba-Sadiya , Hongjing Lu , Sinisa Todorovic , Joyce Chai , and Song-Chun Zhu . 2022. CX-ToM: Counter-factual explanations with theory-of-mind for enhancing human trust in image recognition models. iScience 25, 1 ( 2022 ), 103581. Arjun R. Akula, Keze Wang, Changsong Liu, Sari Saba-Sadiya, Hongjing Lu, Sinisa Todorovic, Joyce Chai, and Song-Chun Zhu. 2022. CX-ToM: Counter-factual explanations with theory-of-mind for enhancing human trust in image recognition models. iScience 25, 1 (2022), 103581.

2. GAM

3. D. Bau , B. Zhou , A. Khosla , A. Oliva , and A. Torralba . 2017 . Network Dissection: Quantifying Interpretability of Deep Visual Representations. In CVPR. 3319--3327. D. Bau, B. Zhou, A. Khosla, A. Oliva, and A. Torralba. 2017. Network Dissection: Quantifying Interpretability of Deep Visual Representations. In CVPR. 3319--3327.

4. D. Bau , J.-Y. Zhu , H. Strobelt , A. Lapedriza , B. Zhou , and A. Torralba . 2020. Understanding the role of individual units in a deep neural network . Proceedings of the National Academy of Sciences ( 2020 ), 30071--30078. D. Bau, J.-Y. Zhu, H. Strobelt, A. Lapedriza, B. Zhou, and A. Torralba. 2020. Understanding the role of individual units in a deep neural network. Proceedings of the National Academy of Sciences (2020), 30071--30078.

5. L. Breiman J. H. Friedman R. A. Olshen and C. J. Stone. 1984. Classification and Regression Trees. Wadsworth and Brooks Monterey CA. L. Breiman J. H. Friedman R. A. Olshen and C. J. Stone. 1984. Classification and Regression Trees. Wadsworth and Brooks Monterey CA.