Multilevel Wavelet Decomposition Network for Interpretable Time Series Analysis

This repository is code release for time series classification task of our recent KDD2018 paper.

The code is based on Tensorflow and other basic python packages.

We only tested the code on Ubuntu 16.04 with python 2.7 and TensorFLow 1.2, a GPU with at least 4GB memory is recommanded.

Usage

Before running the code, modify the line 65 of train.py to let the DATA_ROOT points to the path to UCR dataset on your machine.

To start a demo on training yoga data of UCR dataset, simply run the following command:

python train.py --gpu 0 --log_dir log_demo_yoga

The code is designed to automatically save the best model parameters.

You may try other args by adding them to the command, for details please refer to:

python train.py --help

Citing

If you find our work is helpful for your research, please kindly consider citing our paper as well.

@inproceedings{Wang:2018:MWD:3219819.3220060,
 author = {Wang, Jingyuan and Wang, Ze and Li, Jianfeng and Wu, Junjie},
 title = {Multilevel Wavelet Decomposition Network for Interpretable Time Series Analysis},
 booktitle = {Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery \&\#38; Data Mining},
 series = {KDD '18},
 year = {2018},
 isbn = {978-1-4503-5552-0},
 location = {London, United Kingdom},
 pages = {2437--2446},
 numpages = {10},
 url = {http://doi.acm.org/10.1145/3219819.3220060},
 doi = {10.1145/3219819.3220060},
 acmid = {3220060},
 publisher = {ACM},
 address = {New York, NY, USA},
 keywords = {epidemic propagation, intracity epidemic control and prevention, metapopulation, network inference},
}