This repository contains the code to reproduce the experiments of

@article{Hofer19b,
    author    = {C.~Hofer, R.~Kwitt, and M.~Niethammer},
    title     = {Learning Representations of Persistence Barcdoes},
    booktitle = {JMLR},
    year      = {2019}}

The core folder contains some utility code while the actual training/testing code is in the top-level jupyter notebooks, which are named after the corresponding datasets.

The setup was tested with the following system configuration:

• Ubuntu 18.04.2 LTS
• CUDA 10.1 (driver version 418.87.00)
• Anaconda (Python 3.7)
• PyTorch 1.4

In the following, we assume that we work in /tmp (obviously, you have to change this to reflect your choice and using /tmp is, of course, not the best choice :).

1. Get the Anaconda installer and install Anaconda (in /tmp/anaconda3) using

cd /tmp/
wget https://repo.anaconda.com/archive/Anaconda3-2019.10-Linux-x86_64.sh
bash Anaconda3-2019.10-Linux-x86_64.sh
# specify /tmp/anconda3 as your installation path
source /tmp/anaconda3/bin/activate

2. Install PyTorch (v1.4)

conda install pytorch torchvision cudatoolkit=10.1 -c pytorch

3. Install torchph.

pip install git+https://github.com/c-hofer/torchph.git@0.0.0

4. Clone this GitHub repository.

cd /tmp/
git clone https://github.com/c-hofer/jmlr_2019.git --recurse-submodules

5. Download data

All data can be downloaded here. Unzip the ZIP file using unzip

cd /tmp/jmlr_2019/core
unzip jmlr2019_datasets.zip

This should create a folder datasets in /tmp/jmlr_2019/core/.

6. Start jupyter notebook server in repository folder.

cd /tmp/jmlr_2019
jupyter notebook