Talk by Prof. Yasser Roudi (CSN Virtual Seminar)

We hereby announce the next talk in the 'Computational and Systems Neuroscience Virtual Seminar' in short: 'CSN Virtual Seminar'

Relevance for neural data analysis

In this talk, we first introduce the general concept of “relevance” as a quantitative measure of informativeness of samples and representations (Marsili and Roudi, 2021) and discuss how it naturally leads to the definition of maximally informative samples of data and optimal learning machines. We will then discuss how relevance can be used for identifying relevant neurons in a population without having access to any recorded external covariates. We will discuss how it correlates with other commonly used measures in neural data analysis and show that it can indeed identify neurons that are known to be relevant and informative (Cubero et al 2020).

Prof. Yasser Roudi
Kavli Institute for Systems Neuroscience