Neural Network Attributions: A Causal Perspective
Aditya Chattopadhyay1 Piyushi Manupriya2 Anirban Sarkar2 Vineeth N Balasubramanian2
1Johns Hopkins University 2IIT Hyderabad
In ICML 2019
Paper | Code | Slides | Citations | Media Coverage
Abstract
We propose a new attribution method for neural networks developed using first principles of causality (to the best of our knowledge, the first such). The neural network architecture is viewed as a Structural Causal Model, and a methodology to compute the causal effect of each feature on the output is presented. With reasonable assumptions on the causal structure of the input data,we propose algorithms to efficiently compute the causal effects, as well as scale the approach to data with large dimensionality. We also show how this method can be used for recurrent neural networks.We report experimental results on both simulated and real datasets showcasing the promise and usefulness of the proposed algorithm.
ICML Presentation
Some results
ACE on NASA dataset
ACE on MNIST dataset
Citation
Aditya Chattopadhyay, Piyushi Manupriya, Anirban Sarkar, Vineeth N Balasubramanian. "Neural Network Attributions: A Causal Perspective", in International Conference on Machine Learning (ICML), 2019.
Bibtex
Acknowledgement
We are grateful to the Ministry of Human Resource Development, India; Department of Science and Technology, India;as well as Honeywell India for the financial support of this project through the UAY program. We thank the anonymous reviewers for their valuable feedback that helped improvethe presentation of this work.