The Convergence Rate of Neural Networks for Learned Functions of Different Frequencies

Ronen Basri, David Jacobs, Yoni Kasten, Shira Kritchman (NeurIPS'19)

MATLAB Mex Version 1.0 (2019-02-25) Copyright (C) Yoni Kasten and Shira Kritchman, Weizmann Institute, 2019. Licensed for noncommercial research use only.

Background

This code generates the graphs of the paper.

For more information see:

@article{basri2019convergence,
  title={The Convergence Rate of Neural Networks for Learned Functions of Different Frequencies},
  author={Basri, Ronen and Jacobs, David and Kasten, Yoni and Kritchman, Shira},
  journal={arXiv preprint arXiv:1906.00425},
  year={2019}
}

[arXiv] Please cite these paper if you use this code in an academic publication.

Use

The shallow neural network experiments run in Matlab. The deep neural network experiments run in Python (with PyTorch). Code for all other plots is in Python as well, some of it is in .ipynb format (Jupyter Notebook) and allows for live plotting. The plots generation code is written in Matlab.

Our results can be downloaded from here

For more details please see the specific readme file of each experiment.

License

This software is provided under the provisions of the Lesser GNU Public License (LGPL). see: http://www.gnu.org/copyleft/lesser.html.

This software can be used only for research purposes, you should cite the aforementioned papers in any resulting publication.

The Software is provided "as is", without warranty of any kind.

Version History

• Version 1.0 (2019-10-25) Initial Release