# Modern evolutionary algorithms in computational neuroscience: tools to parameterize, explore model properties & design model structures

A workshop at CNS 2012

Organized by *B. Girard, D. Sheynikhovich, S. Doncieux, J.-B. Mouret & A. Arleo*

Informations: evoneuro_cns_workshop@isir.upmc.fr

In the last 10 years, evolutionary algorithms (EA) have been occasionally used as tools to help parameterizing computational models of the brain. As models grow more and more in complexity, manually adjusting parameters become unreasonable, while automatic approaches, like EA, can provide acceptable solutions.

Modern EA can also help computational neuroscience beyond optimization: the use of multiple-‐objectives EA (MOEA) allows to find multiple trade-‐off solutions to the studied problem, revealing intrinsic properties of the problem itself, and thus actively participating in the investigation process. Modern generative EA can also be used to fully generate the structure of brain networks models based on known anatomical and electrophysiological data and thus directly participate in the model design.

The goal of this workshop is to gather neuroscientists around the current use of EA in computational neuroscience, to advertise the possibilities of this approach, as well as to discuss the emerging and future applications of EA in our field.

## Program

9:00 - 9:30 : B. Girard - Introduction

9:30 - 10:15 : A. Korngreen - Optimizing ion channel models using a parallel genetic algorithm on graphical processors.

*coffee break*

10:30 - 11:15 : D. Jaeger - Using particle swarm evolutionary algorithm to improve ion channel kinetics.

11:15 - 12:00 : J.-C. Quinton - Tuning and learning with evolutionary methods: anticipatory representations and dynamic neural fields.

*lunch break*

13:30 - 14:15 : J.-B. Mouret - Recent advances in evolutionary algorithms with examples from neuro-evolution.

14:15 - 14:40 : E. Lynch - An adaptive exponential integrate-and-fire based model of neural encoding optimized using a parallel genetic algorithm on graphical processing units.

14:40 - 15:05 : C. Ollion - Open-ended evolution of neural models exploiting behavioral consistency

*coffee break*

15:20 - 16:05 : A.J. Nevado Holgado - Multiobjective evolutionary algorithms to fit realistic computational models of brain networks to extensive experimental data.

16:05 - 16:50 : J. Liénard - Integration of detailed primate anatomical and electrophysiological data in a model of the basal ganglia using multi-objective evolutionary algorithms.

16:50 - 17:30 : discussion

## Invited speakers

A. Korngreen (Bar-Ilan University, Israel)

**Optimizing ion channel models using a parallel genetic algorithm on graphical processors**.D. Jaeger (Emory University, USA)

**Using particle swarm evolutionary algorithm to improve ion channel kinetics**.J.-C. Quinton (Pascal Institute / Polytech Clermont-Ferrand, France)

**Tuning and learning with evolutionary methods: anticipatory representations and dynamic neural fields**.A. J. Nevado-Holgado (University of Oxford, UK).

**Multiobjective evolutionary algorithms to fit realistic computational models of brain networks to extensive experimental data**.J.-B. Mouret (Université Pierre & Marie Curie / CNRS, France)

**Recent advances in evolutionary algorithms with examples from neuro-evolution**.J. Liénard (Université Pierre & Marie Curie / CNRS, France)

**Integration of detailed primate anatomical and electrophysiological data in a model of the basal ganglia using multi-objective evolutionary algorithms**.