Neural Egg Seperation

This repository contains demo code for the paper Neural separation of observed and unobserved distributions (ICML 2019).

Please cite the papers below if you make use of the software.

Prerequisites

The following packages are required:

torch
torchvision
numpy 
skimage
museval (for audio)

Usage

First, download speech and shoes/bags training sets from Google drive Store speech train data in data/speech, bags in data/bags, and shoes in data/shoes.

Run

python train_audio.py

Or

python train_images.py

Publications

@inproceedings{halperin2019neural,
  title={Neural separation of observed and unobserved distributions},
  author={Halperin, Tavi and Ephrat, Ariel and Hoshen, Yedid},
  booktitle={Proceedings of the 36th International Conference on Machine Learning(ICML)},
  year={2019},
}