Credit Assignment in Genetic Learning Systems

Abstract

Credit assignment problems arise when long sequences of rules fire between successive external rewards. Two distinct approaches to rule learning with genetic algorithms have been developed, each approach offering a useful solution to a different level of the credit assignment problem. We present a system, called RUDI, that combines features from both approaches. Experimental results are presented that support the hypothesis that multiple levels of credit assignment can improve the performance of rule learning systems based on genetic algorithms.

Cite

Text

Grefenstette. "Credit Assignment in Genetic Learning Systems." AAAI Conference on Artificial Intelligence, 1988.

Markdown

[Grefenstette. "Credit Assignment in Genetic Learning Systems." AAAI Conference on Artificial Intelligence, 1988.](https://mlanthology.org/aaai/1988/grefenstette1988aaai-credit/)

BibTeX

@inproceedings{grefenstette1988aaai-credit,
  title     = {{Credit Assignment in Genetic Learning Systems}},
  author    = {Grefenstette, John J.},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {1988},
  pages     = {596-600},
  url       = {https://mlanthology.org/aaai/1988/grefenstette1988aaai-credit/}
}