Okay, I think I should accept that I am probably outdated or misinformed about the things one can do with neural nets. Re, randomforests, I was referring to the Leo Breiman implementation on the univ berkeley website. Andy Liaw of Merck also maintains the R version of Leo Breiman's algorithm (http://cran.r-project.org/web/packages/randomForest/index.ht...).
I know neural nets have been successfully used for really compute intensive applications like face recognition or vehicle autopilot, but for applications like poker or predicting car sales or data mining used by walmart etc, there has been increasing use of simpler regression based / decision tree based approaches. There is an impression in the machine learning community that NN is like a black box which is probably the reason for its falling popularity if that is the case.
