This code is an attempt to help reproduce results from the paper "Time Series Alignment with Global Invariances" (https://openreview.net/forum?id=JXCH5N4Ujy).
This code is Python3 code and relies on the following libraries:
tslearn>=0.3
numba
geoopt
torch
matplotlib
numpy
scipy
imageio
Dependencies can be installed via:
pip install -r requirements.txt
(maybe the torch
dependency will raise an error and an alternative install
method will be prompted).
Also, the base folder of this code (path to this README.md
file) should be
added to your Python path for dtw_gi
package to work properly:
export PYTHONPATH=$PYTHONPATH:path/to/base/folder
For figures 1 to 5, notebooks are provided that generate the figures: