George H. John's Publications

This is a full list of my publications, ordered roughly according to my own personal preference, which is mostly reverse-chronological. Some unpublished working papers are at the bottom of the list. Click here for help on viewing and printing the files. If you have any comments or questions, you can mail me a note.

[View the Ad] [Get a Copy] [Abstract & Table of Contents]: George H. John. Enhancements to the Data Mining Process, PhD Thesis, Computer Science Department, School of Engineering, Stanford University, 194pp., March 1997.
[Abstract] [Paper]: Ron Kohavi and George H. John. The Wrapper Approach. In H. Liu and H. Motoda, editors, Feature Selection for Knowledge Discovery in Databases. 1998. Springer-Verlag.
[Abstract] [Paper]: George H. John and Pat Langley. Estimating continuous distributions in Bayesian classifiers. In P. Besnard and S. Hanks, editors, Eleventh Annual Conference on Uncertainty in Artificial Intelligence, pages 338--345, San Francisco, 1995. Morgan Kaufmann Publishers.
[Abstract] [Paper]: Ron Kohavi and George H. John. Wrappers for feature subset selection. Artificial Intelligence Journal, Forthcoming. Special Issue on Relevance edited by R. Greiner, J. Pearl and D. Subramanian. [The published version is slightly updated]
[Abstract] [Paper]: George H. John and Brian Lent. SIPping from the data firehose. In D. Heckerman, H. Manilla, and D. Pregibon, editors, Third International Conference on Knowledge Discovery and Data Mining, pages 199--202, Menlo Park, CA, 1997. AAAI Press.
[Abstract] [Paper]: Ron Kohavi and George H. John. Automatic parameter selection by minimizing estimated error. In A. Prieditis and S. Russell, editors, Machine Learning: Proceedings of the Twelfth International Conference, pages 304--312, San Francisco, 1995. Morgan Kaufmann Publishers.
[Abstract] [Paper]: George H. John and Peter Miller. Building long/short portfolios using rule induction. In Computational Intelligence in Financial Engineering, pages 134--140, Piscataway, NJ, 1996. IEEE Press.
[Abstract] [Paper (zipped)]: Vida S. Tigrani and George H. John. Data Mining and Statistics in Medicine: An Application in Prostate Cancer Detection. In JSM98: In JSM98,the Proceedings of the Joint Statistical Meetings, Section on Physical and Engineering Sciences. 1998. American Statistical Association.
[Soft copy not available. See following paper.]: George H. John, Peter Miller, and Randy Kerber. Stock selection using rule induction. IEEE Expert, 11(5):52--58, October 1996.
[Abstract] [Paper]: George H. John, Peter Miller, and Randy Kerber. Stock selection using Recon. In Y. Abu-Mostafa, J. Moody, P. Refenes, and A. Weigend, editors, Neural Networks in Financial Engineering, pages 303--316. World Scientific, London, 1996.
[Abstract] [Paper]: George H. John. Robust decision trees: Removing outliers in databases. In U. M. Fayyad and R. Uthurusamy, editors, First International Conference on Knowledge Discovery and Data Mining, pages 174--179, Menlo Park, CA, 1995. AAAI Press.
[Abstract] [Paper]: George H. John, Ron Kohavi, and Karl Pfleger. Irrelevant features and the subset selection problem. In H. Hirsh and W. Cohen, editors, Machine Learning: Proceedings of the Eleventh International Conference, pages 121--129, San Francisco, 1994. Morgan Kaufmann Publishers.
[Abstract] [Paper (earlier version)]: George H. John. Robust linear discriminant trees. In D. Fisher and H. Lenz, editors, Learning From Data: Artificial Intelligence and Statistics V, Lecture Notes in Statistics, Chapter 36, pages 375--385. Springer-Verlag, New York, 1996.
[Abstract] [Paper]: George H. John and Ying Zhao. Mortgage data mining. In Computational Intelligence in Financial Engineering, pages 232--236, Piscataway, NJ, March 1997. IEEE Press.
[Abstract] [Paper] [Longer Version]: Ron Kohavi, George John, Richard Long, David Manley, and Karl Pfleger. MLC++: a machine learning library in C++. In Sixth National Conference on Tools with Artificial Intelligence, pages 740--743. IEEE Press, 1994.
[Handouts]: George H. John. CS221 Course Handouts: Introduction to Artificial Intelligence. 151 pp. Course given at Stanford University, June-August 1993.
[Abstract] [Paper]: George H. John. Finding multivariate splits in decision trees using function optimization. In Proceedings of the Tenth National Conference on Artificial Intelligence, page 1463, Menlo Park, CA, 1994. AAAI Press.
[Abstract] [Paper]: George H. John. When the best move isn't optimal: Q-learning with exploration. In Proceedings, Tenth National Conference on Artificial Intelligence, page 1464, Menlo Park, CA, 1994. AAAI Press.
[Abstract] [Paper]: George H. John. Cross-validated C4.5: Using error estimation for automatic parameter selection. Technical Report STAN-CS-TN-94-12, Computer Science Department, Stanford University, October 1994.
[Abstract] [Paper]: George H. John. Cascade correlation: Derivation of a more numerically stable update rule. In International Conference on Neural Networks, pages 1126--1129, Perth, W. Australia, 1995. IEEE Press.
[Abstract] [Paper]: George H. John and Pat Langley. Static versus dynamic sampling for data mining. In E. Simoudis, J-W. Han, and U. Fayyad, editors, Proceedings, Second International Conference on Knowledge Discovery and Data Mining, pages 367--370, Menlo Park, CA, 1996. AAAI Press.
[Paper]: George H. John. Annotated bibliography of linear discriminant decision trees. Posted to the comp.ai.neural-nets Newsgroup, May 1994. Available from ftp/dtbib.txt
[Abstract]: Evangelos Simoudis, George John, Randy Kerber, Brian Livezey, and Peter Miller. Developing customer vulnerability models using data mining techniques. In International Symposium on Intelligent Data Analysis, 1995.
[Paper]: George H. John. Upgrade your Macintosh IIsi for cheap! BMUG (Bay Area Macintosh Users Group) Newsletter, Spring, pages 152--154, 1993.

Working Papers by George John

Some of these papers are in submission, others were never submitted or were submitted unsuccessfully, and yet others were submitted successfully but the final version became sufficiently different from the submitted version that I thought it would be worthwhile to make both available. For the most part, either I'm no longer thinking about these problems/algorithms, or I have thought further and the work is published elsewhere. Still, there are some ideas here that might prove useful to someone, so I have made the papers available.

Click on a title to get the postscript file. For each one you can get just the "(Abstract)" which is a text-only web page giving title, author info, the abstract, and citation info for the paper. Look here for published papers.

If you have any comments or questions, you can mail me a note.

When the Best Move Isn't Optimal: Q-learning with Exploration: (Abstract) By George John. May, 1995. Submitted to NIPS*95. This is a longer version of my AAAI student abstract in 1994 with the same title.
Geometry-Based Learning Algorithms: (Abstract) By George John. March, 1993. I think there are some reasonable ideas in here (namely, focusing on geometric and topological properties of learned representations in machine learning), but the paper is a little confused. I'm interested in implementing the convex hull learning algorithm described in this paper, or hearing about someone else who's already done it.
Robust Soft-Entropy Neural Network Trees: (Abstract) By George John. Submitted to Advances in Neural Information Processing Systems 7 (Similar to the "Robust Linear Discriminant Trees" paper on my main page, but includes some info on using neural networks instead of linear discriminants. Not well organized.)
Robust Linear Discriminant Trees: (Abstract) By George John. (Extended Abstract, paper forthcoming) Accepted to Artificial Intelligence & Statistics. (Actually, the paper has come forth. See the paper of the same title on my publications page.)