Skip to content

MATLAB code for training and simulating the chemical-based image classification network detailed in "Leveraging autocatalytic reactions for chemical-domain image classification"

License

Notifications You must be signed in to change notification settings

Chris3Arcadia/AutocatalyticWTA

Repository files navigation

Leveraging autocatalytic reactions for chemical-domain image classification

license

MATLAB code for training and simulating the chemical-based image classification network detailed in our paper:

"Leveraging autocatalytic reactions for chemical-domain image classification" by Christopher E. Arcadia, Amanda Dombroski, Kady Oakley, Shui Ling Chen, Hokchhay Tann, Christopher Rose, Eunsuk Kim, Sherief Reda, Brenda M. Rubenstein, and Jacob K. Rosenstein. Chem. Sci. (2021). https://doi.org/10.1039/D0SC05860B

Requirements

To run this code, the following are required:

Usage

  1. Download the CalTech 101 data set

  2. update the field database_path in the configure_options.m file with the database location.

  3. Review and optionally edit the other program settings in the configure_options.m file.

  4. Run the script main.m in MATLAB to train and simulate the network.

Also included is a script to analyze the images in the provided dataset (analyze_dataset.m) as well as a script to run multiple classification tests (test_multiple_classifiers.m).

Example

Running the main.m script, after changing the default classes and epoch count to:

option.classes = {'revolver','lamp','mandolin','headphone','umbrella','helicopter','pyramid','chair','saxophone'};
option.epochs = 900; 

results in the following weight maps:

and class predictions:

Cite

Please cite our paper if you use this code in your own work:

@article{arcadia2021,
  author = {Arcadia, Christopher E and Dombroski, Amanda and Oakley, Kady and Chen, Shui Ling and Tann, Hokchhay and Rose, Christopher and Kim, Eunsuk and Reda, Sherief and Rubenstein, Brenda M and Rosenstein, Jacob K},
  title = {Leveraging autocatalytic reactions for chemical-domain image classification},
  journal = {Chem. Sci.},
  year = {2021},
  publisher = {The Royal Society of Chemistry},
  doi = {10.1039/D0SC05860B},
  url = {http://dx.doi.org/10.1039/D0SC05860B},
}

Related Work

This work has been developed by the Molecular Informatics Team at Brown University (Providence, RI, USA). Visit our website for related projects and publications.

License

Copyright © 2021, Christopher E. Arcadia

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

About

MATLAB code for training and simulating the chemical-based image classification network detailed in "Leveraging autocatalytic reactions for chemical-domain image classification"

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages