alex-delalande/dl_tda

Deep Learning with Topological Signatures

This repository contains the code to reproduce the results of the following paper:

@inproceedings{Hofer17c,
  author    = {C.~Hofer and R.~Kwitt and M.~Niethammer and A.~Uhl},
  title     = {Deep Learning with Topological Signatures},
  booktitle = {NIPS},
  year      = 2017}

#Read me first:

1. The intent of this repository is to reproduce the results of Hofer17c. If you are looking for code optimized for reuse allow me to refer you to chofer_torchex and tda-toolkit.
2. I have tested the code on ubuntu 14.04 and 16.04 system setups. Since this is more or less a two man show testing was not as intensive as it could have been. So if you use the code I consider you as beta tester :).

Installation

1. Ensure PyTorch is installed properly. (During developement I used PyTorch 0.2)

2. If you want to calculate the persistence diagrams yourself make sure the tda-toolkit submodule is configured properly, see tda-toolkit for how to do this.

3. Clone the repo with the --recursive flag set (otherwise the submodules won't be cloned). If you want to install the submodules manually you can omit the flag.

Usage

In order to reproduce the results for a specific data set just run the corresponding scripts in the root folder of the repo, i.e.,

cd /dir/to/nips2017
python animal.py