Tom Fawcett
contact:
tom.fawcett at gmail dot com
I am affiliated with the Stanford Computational Learning Laboratory, a part of the Stanford Center for the Study of Language and Information (CSLI).
Sabbatical
For the past year I've been on sabbatical, by which I mean I've taken the year off to pursue some research interests unrelated to what I've done in my work.
- I've become interested in swarm and collective intelligence: the ways in which collections or communities can learn, adapt and exhibit intelligence. I'm co-editing a special issue of the Machine Learning Journal on swarm intelligence, which will appear this year.
- I been working on the topic of classification with cellular automata, which I've published in SigKDD Explorations. I'm extending and expanding this work for a paper to submit to ICDM in a June.
- I wrote a rule learning system for maximizing the area under the ROC curve (AUC). I published a KDDM paper on it, and now I'm in the process of reimplementing it in Python so I can release it as open-source.
- I'm working on an implementation of my cellular automata work in a new nanotechnology called Quantum Dot Cellular Automata. I've been using the QCADesigner simulator from UBC. I estimate I'm about 80% finished with the necessary circuitry. (I'm also learning about quantum mechanics so I can understand some of the basic physics of this technology).
Professional Service
I am an action editor of the Machine Learning Journal. I'm was a program chair of ICML-03 (The Twentieth International Conference on Machine Learning).
- I guest edited a special issue of Machine Learning journal on Data Mining Lessons Learned.
- I am currently guest editing (with Bart Baesens and David Marten) a special issue of Machine Learning on Swarm Intelligence for Knowledge Discovery in Data. The Call for Papers is here.
CV, Publications, etc.
- My publications on machine learning, data mining and knowledge discovery.
-
My CV in PDF format.
- Software, in Perl, related to machine learning and data mining.
-
The ROC Convex Hull: Papers, viewgraphs, bibliography and a program.
-
Books for machine learning people which are not machine learning books. Machine learning and data mining have become very narrow in the last five years or so. These are some books I recommend to remind people that the phenomena we call learning are actually much broader, richer and more fascinating than the few types we study.
-
Bibliographies I maintain:
-
Spam and spam filtering. This bibliography contains the references used in my spam paper plus some others that I've been accumulating.
-
Feature engineering and constructive induction. This is available from the Collection of Computer Science Bibliographies, which has very nice search and presentation facilities.
-
Fraud detection. This is available through the web interface of the Collection of Computer Science Bibliographies, and as a raw BibTex file.
-
-
Workshops I've organized:
-
Constructive Induction and Change of Representation (MLC-94)
-
AI Approaches to Fraud Detection and Risk Management(AAAI-97)
-
AI Approaches to Time-Series Problems (AAAI/ICML-98)
-
Data Mining Lessons Learned (ICML-2002)
-
Spam
One of my interests is the problem of spam detection and filtering, as well as general email classification.
-
I studied spam and spam filtering for a while, and then wrote a paper: "In vivo" spam filtering: A challenge problem for data mining. The audience is data mining researchers, but I recommend the paper to anyone interested in studying the problem. The real spam filtering problem has a lot of interesting aspects that most people ignore.
-
I was on the program committee of the First Conference on Email and Anti-Spam (CEAS-04)
-
I was program chair of the Second Conference on Email and Anti-Spam (CEAS-05).
-
I'm on the Technical Advisory Council of Proofpoint, Inc., a spam filtering company.
