ironshim / 1210_dtd_ig

Deep Taylor Decomposition and Integrated Gradients Tutorial

This is the repository for XAI tutorials of KAIST-LG Advanced AI course. This repository provides tutorial code for Deep Taylor Decomposition [1] and Integrated Gradients [2].

Dependencies

• Python 3.7
• Tensorflow 1.14.0
• Tensorboard 1.2.0
• Numpy 1.18.1
• Matplotlib 3.1.3

Usage

Train

$ python main.py --train

Test

$ python main.py --test

Explanation

// Running Deep Taylor Decomposition
$ python main.py --explain --method dtd

// Running Integrated Gradients
$ python main.py --explain --method integrated

Examples

References

[1] Montavon, G., Lapuschkin, S., Binder, A., Samek, W., & Müller, K. R. (2017). Explaining nonlinear classification decisions with deep taylor decomposition. Pattern Recognition, 65, 211-222.

[2] Sundararajan, M., Taly, A., & Yan, Q. (2017). Axiomatic attribution for deep networks. arXiv preprint arXiv:1703.01365.