TimeNet

February 16, 2021 · View on GitHub

Implementation of TimeNet autoencoder for embedding time series. Based on Malhotra et al. (2017) "TimeNet: Pre-trained deep recurrent neural network for time series classification", https://arxiv.org/abs/1706.08838

Addtitionally, this implementation supports features not given in the original paper:

Support for variable length time series
Time series normalization
Variable-size batching with respect to non-uniform lengths of input time series

Implemented using TensorFlow 2.3.0 with Keras API. Experiments performed using time series data from http://www.cs.ucr.edu/~eamonn/time_series_data/.

Usage

Can be run from command line using main.py script. Sample data is provided in data folder