Satisficing and Bounded Optimality

Shlomo Zilberstein. Satisficing and Bounded Optimality. AAAI Spring Symposium on Satisficing Models, Stanford, California, 1998.

Abstract

Since the early days of artificial intelligence there has been a constant search for useful techniques to tackle the computational complexity of decision making. By now, it is widely accepted that optimal decision making is in most cases beyond our reach. Herbert Simon's approach based on satisficing offers a more realistic alternative, but it says little on how to construct satisficing algorithms or systems. In practice, satisficing comes in many different flavors, one of which, bounded optimality, restores a weak form of optimality. This paper demonstrates this form of satisficing in the area of anytime problem-solving and argues that it is a viable approach to formalize the notion of satisficing.

Bibtex entry:

@inproceedings{Zspring98,
  author	= {Shlomo Zilberstein},
  title		= {Satisficing and Bounded Optimality},
  booktitle     = {{AAAI} Spring Symposium on Satisficing Models},
  year		= {1998},
  pages		= {},
  address       = {Stanford, California},
  url		= {http://rbr.cs.umass.edu/shlomo/papers/Zspring98.html}
}

shlomo@cs.umass.edu
UMass Amherst