eeg-b2b-regression

Back-to-Back regression for encoding and decoding analysis

Introduction

There are broadly two categories of EEG analyses: Decoding, f(brain) = stimulus and encoding f(stimulus) = brain. Here, we will try to combine the benefits of both methods based on the analysis-approach of back2back regression, which in some sense encompasses both.

Steps

We use Pluto.jl notebooks to analyse our result.

• We simulate EEG data [DataSimulation.jl], and apply back-2-back regression using regularized(L1, L2, Elastic) and un-regularized solvers with the help of Unfold.jl. We also explore single layer neural network solver.

• We apply back-2-back regression [DataAnalysis.jl] on ground truth data to disentangle the effects of continuous correlated predictors and uncorrelated categorical predictor.

• We generate saliency maps and analyse saliency scores.

Setup

Install dependencies:

julia> # julia REPL
julia> ]  # enter pkg mode
> activate .
> instantiate

To run Pluto

julia> import Pluto;
julia> Pluto.run()

References

Ehinger BV, Dimigen O. 2019. Unfold: an integrated toolbox for overlap correction, non-linear modeling, and regression-based EEG analysis. PeerJ 7:e7838 https://doi.org/10.7717/peerj.7838

Jean-Rémi King, François Charton, David Lopez-Paz, Maxime Oquab, Back-to-back regression: Disentangling the influence of correlated factors from multivariate observations, NeuroImage, Volume 220, 2020, 117028, ISSN 1053-8119, https://doi.org/10.1016/j.neuroimage.2020.117028.(https://www.sciencedirect.com/science/article/pii/S1053811920305140)