Next: Further reading --- game trees
Up: Conventional AI: Search
Previous: Introduction
The OU course will give you all you need to know about breadth-first and
depth-first search and evaluation functions.
You must also know the difference between blind and guided
(heuristic) search. The OU book does introduce these, but
in a way which you may find confusing. To make things
clearer, read chapter 4 of Artificial Intelligence by Winston, and
work through an example of hill-climbing.
It's as well to know that
there are other search methods, such as A*.
It's useful to realise why we need so many different search methods,
and how their characteristics differ. See Winston for this.
- Article on Search
in The Encyclopaedia of Artificial Intelligence,
(RSL: Comp Bd 56)
edited by Shapiro. A handy compare-and-contrast of different methods
for the computer-minded.
It describes depth-first iterative-deepening, a method not mentioned
in my other references. This is useful when you want something that
has the advantages of breadth-first without its massive space overheads.
- Artificial Intelligence, Winston, Chapter 4. Gives
algorithms for the various methods. Since they're all in the same
style, they're useful if you want to compare how a programmer would
solve search problems. Also useful for comparison with GPS next week.
- Artificial Intelligence by Charniak and McDermott: RSL
stacks; Psychology.
Because so many of the problems we discuss are puzzles, you may be left
with the impression that these methods are useless for ``real-world''
problems. Charniak and McDermott give an example of where they can be
useful.
Jocelyn Paine
Tue Jun 3 11:15:49 BST 1997