[Colloq] **Hiring Talk **, Friday, January 30, 10:30am

[Rachel Bates]

College of Computer and Information Science Colloquium

presents
Gregory Piatetsky-Shapiro

  who will speak on:
Explorations in Knowledge Discovery


Friday, January 30, 2004
10:30am
149 Cullinane Hall
Northeastern University



ABSTRACT
It was estimated that 5 million terabytes of new data was created in 2002
alone, with data storage doubling about every 9 months.  To make sense of
this flood of data, we need the science and tools for data mining and
knowledge discovery.

I describe some highlights of my work and research in this field.  One
notable project was Key Findings Reporter (KEFIR), which was applied to a
large health care database. KEFIR used domain knowledge to analyze changes
in all relevant variables, select the important ones, and add expert
recommendations for actions.

Another project was empirical meta-analysis of quality of models for
predicting consumer behavior, which found a surprising formula for the
average lift curve.

I also present my current research on DNA microarray data analysis,
including development of best practice templates for classification and
research on comprehensive microarray testbed for objectively comparing
existing algorithms.

Host:  Karl Lieberherr