This is the code base for the experiments for **Neural Bregman Divergences for Distance Learning**

Each script in the base directory (e.g. fit_bregmnist.py, div_learning.py) runs a separate task using command-line options. As presented in paper:
    - distributional clustering: distribution_learning.py
    - standard Bregman clustering/ranking: div_learning.py
    - Bregman regression: regression.py
    - Bregmnist/Bregcifar: fit_bregmnist(cifar).py
    - deep metric learning: deep_div_learning.py
    - semantic distance: fit_clust_cifar.py
    - overlap distance: fit_cropdist.py
    - graph distances: graph_distance/learn_graph_dist.py

Implementation of Bregman divergence (and other baselines) is in ./divergence

Full paper with Appendix included in repo.
