High-level description. In a methods paper, I discuss how a machine learning method - the Support Vector Machine (SVM), can be applied to electrophysiology recordings of neural activity. SVMs are usually used for (binary) classification, but beyond its use as a classifier, the linear SVM can also be used to study the organisation of high-dimensional datasets such as spike trains emitted by medium sized populations of neurons. How the method can be applied to such data and what do we lear from it is presented in a step-by-step “cookbook” format, where each step of the analysis is commented and discussed. I also expand on when and why described methods are expected to bring insights about neural activity. This paper is oriented towards readers that would like to learn about SVMs. It also fosters reproducibility of results of the main publication.
Author: Veronika Koren
Published in: STAR Protocols, 2021 | to STAR Protocols