Author: Vincent Diepeveen
Date: 01:52:10 07/27/01
Go up one level in this thread
On July 25, 2001 at 22:33:16, Pham Minh Tri wrote: >Machine learning and auto adjusting the evaluation function are big and >interesting fields. > >I know that someones are applying those fields and having good results. For >example to encourage you, Vincent (from his post) has been using Neural Network >and Machine learning technique to improve his strong and famous program - DIEP. >Someothers do not like them because of ineffectiveness. What's this for nonsense? I only could conclude NNs failed bigtime to just find a *normal* parameter set when the number of parameters get big. Also though officially finding better parameters is called 'learning', i do not see it as real learning. Real learning is if you let the machine find the patterns. What i tried was parameter optimizing techniques at an automated way and that failed so miserably that the obvious conclusion was that you need domain specific knowledge to optimize the parameters. More interesting as another going to fail attempt to create something that optimizes parameters is making a genetic algorithm to factorize big numbers. Best regards, Vincent >Good luck and have fun with your project. >Pham
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.