Plenary Talk 2

  • 홈ˆ
  • Program
  • Plenary Talk 2

[Plenary Talk 2]    11.2(Thur) 11:00 - 12:30

Bert Kappen
Radboud Univ. Nijmegen

A control based view on intelligence

▷ Abstract
When comparing how humans learn to how robots learn, there are a number of striking differences: Humans are continuously engaged in autonomous learning of many tasks simultaneously, generating massive amounts of data from a very large number of real-time sensor readings. Instead, robot learning is characterized by the use of a small number of trials (10-100), aims to learn individual tasks, and use limited sensing. As a result, the learned robot behavior poorly generalizes to new tasks and situations.

In this talk, I will introduce a novel approach to modelling intelligent behaviour that integrates control and learning. The key idea is the recently developed path integral control theory, that has the ability to solve large non-linear stochastic control problems in real time. The presentation consist of two parts: In the first part, the path integral control theory is presented and illustrated on some robotic tasks. It is shown, that motor learning relies essentially on stochastic simulation. In the second part, it is argued how this idea can be embedded in a cognitive architecture.

▷ Bio
Bert Kappen is professor of physics and head of the SNN Research group at the Radboud University in Nijmegen, the Netherlands and honorary faculty at UCL’s Gatsby Computational Neuroscience Unit in London. Since 1989, he is conducting research on neural networks, Bayesian machine learning, stochastic control theory and computational neuroscience. His research has made significant contributions to approximate inference in machine learning using methods from statistical physics; he has pioneerd the field of path integral control methods for solving large non-linear stochastic optimal control problems and their relation to statistical physics. He co-founded in 1998 the company Smart Research that commercialises applications of neural networks and machine learning, such as forensic software for DNA matching used by leading forensic institutes in the Netherlands, Australia, Vietnam and Interpol

※ The above schedule is subject to change.

서울시 서초구 방배로 76 (방배동, 머리재빌딩 401호) 우)06704 | (Tel)1588-2728 | (Fax)02-521-1352 | 사업자등록번호 : 114-82-03170 | 대표 : 나연묵

Copyright (c) KIISE. All rights reserved.