Goals

The aim of this course is to model and solve certain complex problems using so-called “collaborative” algorithms. These have the peculiarities of taking an example from nature (genetic algorithms, ant colonies, ..., neural networks) and of using the collective experience of "individuals" (agents) with weak capacities to make one. collective intelligence. For example, neural networks seek to mimic the brain's ability to solve a problem by using the multitude of neurons (each with poor resolving capacity) that make it up. The applications dealt with in progress are varied: character recognition, detection of outlines (in an image), resolution of a poker game (simplified) (or even other games), decoding of a text, search for a path the shortest (Dijkstra, traveling salesman), fault detection, bus allocation and Simultaneous Mapping and Localization by use of robots ...

Programme

  • Algorithmes Génétiques
  • Réseaux de Neurones
  • Déplacement probabiliste et perception en Robotique, SLAM
  • Optimisation par essaim de particules (PSO), colonies de fourmis (ACO)
Course
8h
 
TC
16h
 
PW
8h
 

Responsibles

  • Philippe MICHEL
  • Alexandre SAIDI

Language

French

Keywords

multi-agents, robotics, genetic algorithms, ant colonies, neural networks, slam