Activity of large neural populations can be represented by few independent patterns called “synergies”. Synergies constrain neural dynamics in specific manifolds. Standard dimensionality-reduction techniques have been used to find sets of synergies. Neural spaces may contain distinct manifolds, each associated with specific tasks: is each manifold spanned by its own unique synergies, or do tasks in a class share some synergies that represent common features? We use sparse dictionary-learning approaches to tackle these open issues.

This presentation is part of Minisymposium “MS53 - Dimensionality Reduction Algorithms for Large-Scale Images (2 parts)”
organized by: Salvatore Cuomo (Dept. Mathematics and Applications "Renato Caccioppoli", University of Naples) , Costantinos Siettos (National Technical University of Athens ) , Lucia Russo (Consiglio Nazionale delle Ricerche, Istituto di Ricerche sulla Combustione) .

Authors:

Francesco Donnarumma (Institute of Cognitive Sciences and Technologies, Rome)

Roberto Prevete (DIETI University of Naples Federico II)

Keywords:

dictionary learning, machine learning, neural population codes