Apologies for not quoting Don Dailey's text on Borda voting -- yahoo is doing
something truly awful with quoted text, for some reason.
David Doshay has mentioned both on-list and in offline conversation that one
hurdle faced by multi-agent intelligences is the varying scales used to rank
moves. UCT-MC uses probability of win; Gnugo uses a measure of expected
territory, plus some other measures. There are many other diverse ways to rank
moves by various criteria. Different programs measure different things by
different scales; combining these is a difficult problem.
Borda voting provides a neat way of combining various scales -- it relies on a
weaker constraint than commensurability, that of ordering. One agent prefers
move A to B, and B to C; another prefers B to A and A to C; and so forth. We
sidestep the questions of scale and combinations of totally different
quantities - probability versus territory versus whatever other qualities might
be measured.
We can imagine ways of indicating greater degrees of emphasis -- if one votes
for only the top three moves on the board, this indicates a degree of urgency
compared to an agent which thinks twenty moves are interesting enough to rank.
If we trust one agent more than others, we might weight the votes accordingly.
That weighting might itself be trainable.
Terry McIntyre <[EMAIL PROTECTED]>
“Wherever is found what is called a paternal government, there is found state
education. It has been discovered that the best way to insure implicit
obedience is to commence tyranny in the nursery.”
Benjamin Disraeli, Speech in the House of Commons [June 15, 1874]
____________________________________________________________________________________
Looking for last minute shopping deals?
Find them fast with Yahoo! Search.
http://tools.search.yahoo.com/newsearch/category.php?category=shopping
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
