variational Latent Gaussian Process

Introduction

This repo contains the implementation of variational Latent Gaussian Process (vLGP) (arXiv) (video) by Yuan Zhao (yuan.zhao@stonybrook.edu) and Il Memming Park (memming.park@stonybrook.edu). It has been developed with the goal of recovering low-dimensional dynamics from neural population recordings.

Installation

pip

pip install git+https://github.com/catniplab/vlgp.git

Usage

The main entry is vlgp.fit. The fit function requires two arguments trials and n_factors. The former is expected as a list of dictionaries, each of which stores on trial and at least contains a identifier ID and the observation y in the shape of (bin, channel). The later specifies the number of factors (latent processes).

result = vlgp.fit(
    trials,       # list of dictionaries
    n_factors=3,  # dimensionality
)

The fit function returns a dictionary of trials, params and config as the fitted model.

Please see the tutorial for details.

Citation

@Article{Zhao2017,
  author    = {Yuan Zhao and Il Memming Park},
  title     = {Variational Latent Gaussian Process for Recovering Single-Trial Dynamics from Population Spike Trains},
  journal   = {Neural Computation},
  year      = {2017},
  volume    = {29},
  number    = {5},
  pages     = {1293--1316},
  month     = {may},
  doi       = {10.1162/neco_a_00953},
  publisher = {{MIT} Press - Journals},
}

Changes

2018

• New uniform data structure
• Support trials of unequal duration
• Faster
• Use NumPy data format

2017

• New fit function now only requires observation and the number of latent.
• Save snapshots if path is passed to fit.
• You can access the iterations via callback.