inria-00173241, version 1
Slightly beyond Turing-Computability for studying Genetic Programming
MCU'07 (2007)
Abstract: Inspired by genetic programming (GP), we study iterative algorithms for non-computable tasks and compare them to naive models. This framework justifies many practical standard tricks from GP and also provides complexity lower-bounds which justify the computational cost of GP thanks to the use of Kolmogorov's complexity in bounded time.
- 1: TAO (INRIA Futurs)
- INRIA – CNRS : UMR8623 – Université Paris XI - Paris Sud
- Domain : Mathematics/Optimization and Control
- Keywords : Genetic Programming – computability – complexity
- inria-00173241, version 1
- http://hal.inria.fr/inria-00173241
- oai:hal.inria.fr:inria-00173241
- From: Olivier Teytaud
- Submitted on: Wednesday, 19 September 2007 14:16:47
- Updated on: Wednesday, 19 September 2007 14:18:53
Associated documents
Export