441 search results - page 3 / 89 » Convergence of Non-Elitist Strategies |
ECML
2005
Springer
14 years 29 days ago
2005
Springer
Abstract. While direct, model-free reinforcement learning often performs better than model-based approaches in practice, only the latter have yet supported theoretical guarantees f...
EVOW
2005
Springer
14 years 29 days ago
2005
Springer
This paper explores strategies for slowing the onset of convergence in an evolving population of agents. The strategies include the emergent maintenance of separate agent sub-popul...
AE
2005
Springer
14 years 1 months ago
2005
Springer
Randomization is an efficient tool for global optimization. We here define a method which keeps : – the order 0 of evolutionary algorithms (no gradient) ; – the stochastic as...
ICML
2001
IEEE
14 years 8 months ago
2001
IEEE
As multiagent environments become more prevalent we need to understand how this changes the agent-based paradigm. One aspect that is heavily affected by the presence of multiple a...
MOC
2002
13 years 7 months ago
2002
This paper shows that for unitary Hessenberg matrices the QR algorithm, with (an exceptional initial-value modification of) the Wilkinson shift, gives global convergence; moreover,...