Skip to content

[WIP] Collection of models from various projects organized for general use

License

Notifications You must be signed in to change notification settings

neuralsignal/scidoggo

Repository files navigation

WIP: A collection of shallow and deep models

This project is still under development. This project contains a selection of models and data tools I have created, modified, and used for different projects.

Models

  • Selectivity Model: model that non-linearly integrates input features to produce concave and convex isoresponse surfaces
  • Partial Least Squares: modification of sklearn.pls models to allow for optional bias removal and orthogonalization of rotations and loadings
  • RbfModel: modified and sklearn-compatible version of the scipy.interpolate.RBFInterpolator model
  • Collection of probabilistic circuit model designs using pytorch and pyro
    • These models allow you to model circuits and incorporating anatomical constraints as priors
  • Deep implicit circuit models designed using pytorch and lightning
    • The models allow you to model constrained circuits assuming the observed responses are at steady-state
  • Rank1PlusSparse: linear model with a rank one constraint and an added sparse weight matrix
  • RankConstraint: linear model with a rank constraint (different approach to PLS)
  • Tikhonov regression
  • TwoLayerEncodingModel: two layer encoding model with different objective functions

Acknowledgments

This package was created with the help of the scikit-learn templating tool: https://github.com/scikit-learn-contrib/project-template

About

[WIP] Collection of models from various projects organized for general use

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages