> support-vector machines, neural networks have been considered completely > obsolete in the machine-learning community. From a marketing point of > view, it is not a good idea to do research on neural networks nowadays. > You must give your system another name.
That seems to be the case in the academic community, but people are still using and promoting neural nets to do stuff. Here is something I stumbled upon the other data, where a neural net is being used to "provide helpful situational awareness" in games: http://games.slashdot.org/story/09/05/14/0447244/Java-Program-Uses-Neural-Networks-To-Monitor-Games The underlying library is open source: http://neuroph.sourceforge.net/ A key feature of go is that small differences can dramatically change the result, and neural nets (and similar machine learning) tend to do badly in that kind of situation. But MCTS is a search mechanism that positively thrives on inexact inputs. Using nets to provide higher-level-but-a-bit-rough information to MCTS seems like it good be very productive. (Very interesting to hear Remi say his pattern-learning work could be described as a neural net; I'd never thought of it in that way.) Darren -- Darren Cook, Software Researcher/Developer http://dcook.org/gobet/ (Shodan Go Bet - who will win?) http://dcook.org/mlsn/ (Multilingual open source semantic network) http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
