http:^^www-ml.cs.umass.edu^~utgoff^
来自「This data set contains WWW-pages collect」· EDU^~UTGOFF^ 代码 · 共 109 行
EDU^~UTGOFF^
109 行
Date: Thu, 21 Nov 1996 20:42:21 GMTServer: NCSA/1.5Content-type: text/htmlLast-modified: Wed, 30 Oct 1996 14:22:35 GMTContent-length: 3650<html><head><title>Paul Utgoff</title></head><body><!WA0><img src="http://www-ml.cs.umass.edu/~utgoff/utgoff.gif" alt="Picture"><h2>Paul Utgoff</h2><p><!WA1><a href=mailto:utgoff@cs.umass.edu>Paul Utgoff</a> is AssociateProfessor in the <!WA2><a href="http://www.cs.umass.edu/">Computer ScienceDepartment</a> at the <!WA3><a href="http://www.umass.edu/">University ofMassachusetts at Amherst</a>, where he directs the Machine LearningLaboratory. </p><p>Surface mail may be sent to Professor Utgoff at: Department of Computer Science,Lederle Graduate Research Center, Box 34610, University of Massachusetts, Amherst, MA 01003-4610.Telephone number is (413) 545-4843, and FAX number is (413) 545-1249. E-Mail addressis <!WA4><a href=mailto:utgoff@cs.umass.edu>utgoff@cs.umass.edu</a>.</p><hr><h3>Research</h3><p>I am interested in various forms of decision tree induction. Imaintain a system called ITI, which is described in a <!WA5><ahref="ftp://ftp.cs.umass.edu/pub/techrept/techreport/1995/UM-CS-1995-018.ps">technical report</a>. The <!WA6><ahref="ftp://ftp.cs.umass.edu/pub/iti/iti-2.8.tar.gz">source code andprogram documentation</a> are also available. More recently, I havebeen developing a test selection metric based on Kolmogorov-Smirnoffdistance, which is also described in a <!WA7><ahref="ftp://ftp.cs.umass.edu/pub/techrept/techreport/1996/UM-CS-1996-003.ps">technical report</a>. </p><p>I am intrigued with problem of how to find good featuresautomatically while learning a value function, particularly in gameplaying. I have been thinking about practical methods for exploring aspace of feature functions. One such approach is described in arecent <!WA8><a href="http://www-ml.cs.umass.edu/~utgoff/publications/96-65.ps">technical report</a>. This is where I'm spending most of my time.</p><p>I have developed an approach to writing game-playing programs thatcompete with one another over the internet. The approach does notrequire a central server, and it allows anyone on the internet to runa contest of player programs located anywhere on the internet. Thisapproach is described in a <!WA9><ahref="ftp://ftp.cs.umass.edu/pub/techrept/techreport/1995/UM-CS-1995-067.ps">technical report</a>. Code distributions are available for <!WA10><ahref="ftp://ftp.cs.umass.edu/pub/internet-games/othello-2.15.tar.gz">Othello</a>, <!WA11><ahref="ftp://ftp.cs.umass.edu/pub/internet-games/hearts-2.14.tar.gz">Hearts</a>, and <!WA12><ahref="ftp://ftp.cs.umass.edu/pub/internet-games/checkers-2.2.tar.gz">Checkers</a>. We have an active Othello group, and a Hearts group ofabout ten people has just formed in conjunction with seminar onAdversary Search being taught Fall 1996. My own principal interest ingame playing pertains to the representational issues in finding goodfeatures, but I am very interested in the search aspects too. </p><p>David Aha maintains a <!WA13><a href="http://www.aic.nrl.navy.mil/~aha/people.html ">list</a> of people withinterests in machine learning and case-based reasoning.</p><h3>Service</h3><p>I am currently serving on the Editorial Board of Machine Learning Journal,and the Organizing Committee of ICML-97.</p><h3>Teaching</h3><p>During Fall 1996 I am teaching the <!WA14><a href="http://edlab-www.cs.umass.edu/cs187/">Data Structures</a> course, anda seminary on <!WA15><a href="http://edlab-www.cs.umass.edu/cs691e/">Adversary Search</a>.</p><h3>Online Publications</h3><p><ul><li> Utgoff, P. E. (1996) <!WA16><a href="http://www-ml.cs.umass.edu/~utgoff/publications/96-65.ps">ELF: An Evaluation Function Learner ThatConstructs Its Own Features</a>, Report TR-96-65, Department of Computer Science, University ofMassachusetts, Amherst, MA 01003.</ul></p><hr><p>Last updated October 23, 1996</p></body></html>⌨️ 快捷键说明
复制代码Ctrl + C
搜索代码Ctrl + F
全屏模式F11
增大字号Ctrl + =
减小字号Ctrl + -
显示快捷键?