Data repository of the paper "Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems"
Creators
- 1. Cornell University & NTT Research
- 2. Cornell University
Description
This data repository contains the information necessary to reproduce the main results of the paper “Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems”. The repository contains both the data required to generate the figures of the main manuscript and the supplemental material, as well as source code for running the experiments conducted in the paper.
The repository is structured around each of the three physical systems (oscillating plate, analog transistor, and second harmonic generation) used to construct a physical neural network (PNN). In each folder, there is a README.txt file that summarizes the structure and content of the corresponding folder. Throughout the whole data repository, we have partitioned the source code for running the experiments from the code required to generate the figures from the experimental data by having separate Jupyter notebooks.
The codes in this repository are presented for the purpose of reproducing the results in the main manuscript and supplementary material. For readers who are interested in developing their own applications of physics-aware training and physical neural networks, we recommend starting with our expandable code and its tutorial examples, available at the Github repository: https://github.com/mcmahon-lab/Physics-Aware-Training.
Finally, the simulations, experiments, and generation of the figures were conducted in Python. Since some parts of the code require specific versions of Python packages to run, the required packages are specified in “requirement.txt”.
Files
analog_transistor_PNN.zip
Files
(8.4 GB)
Name | Size | Download all |
---|---|---|
md5:b169973d78713522377223c570d023e0
|
6.0 GB | Preview Download |
md5:fd0f46153fe36f169ddb9a3ac5ce49fb
|
803.5 MB | Preview Download |
md5:87820f6aa3d68d2dd6c88b182e4670c9
|
1.6 kB | Preview Download |
md5:3bb4d270b41d8bbfe13230f341524d3d
|
10.1 kB | Preview Download |
md5:47e85cac07908288092567ef8747b41a
|
1.5 GB | Preview Download |