Mailing List Archive Mailing List Archive
> A neural network *can* do the job, but with some (unknown) probability of > err. This can be 1. So you are much better off with a proper model in the > first place. > John Don't dismiss nnets so easily; they have been shown to be superior to "proper models" in many situations, especially situations with many variables and potentially unknown factors. Sounds a lot like deco modeling, to me... As for probability of error, a nnet is certainly no worse than a hand-built model; in fact, it may be better, as the statistics of the various learning algorithms for nnets are becoming relatively well understood (you can derrive lots of info about potential error from watching the ststistics during training). -frank -- fhd@pa*.co* | Space is not a lot of points close together; it is 1 212 559 5534 | a lot of distances interlocked. 1 917 992 2248 | -- A. Eddington, 1 718 746 7061 | _Mathematical Theory of Relativity_
Navigate by Author:
[Previous]
[Next]
[Author Search Index]
Navigate by Subject:
[Previous]
[Next]
[Subject Search Index]
[Send Reply] [Send Message with New Topic]
[Search Selection] [Mailing List Home] [Home]