This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.

Abstract Details

Activity Number: 356
Type: Contributed
Date/Time: Tuesday, August 3, 2010 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistical Learning and Data Mining
Abstract - #307455
Title: Outlier Detection Methods for Improving Boosting
Author(s): Waldyn Martinez Cid* and J. Brian Gray+
Companies: The University of Alabama and The University of Alabama
Address: Dept of Info Systems, Statistics, and Mgt Science, Tuscaloosa, AL, 35487-0226,
Keywords: classification ; ensemble model ; linear programming ; machine learning ; noisy data ; predictive modeling
Abstract:

Boosting is a method that combines many fairly inaccurate weak learners into a highly accurate prediction rule. The boosting algorithm repeatedly calls these weak learners, each time using a different weighting of the training examples, specifically assigning more weight to those observations that are inaccurately classified. While AdaBoost (Freund and Schapire 1997) is the best known boosting algorithm, there are many different implementations of the algorithm. Dietterich (2000) and others have shown that boosting places too much emphasis on outliers, which often leads to overfitting of the data. Friedman, Hastie, and Tibshirani (2000) and others have proposed variants of the AdaBoost algorithm for improving the resistance of boosting to outliers. In this paper, we describe new methods for improving the performance of boosting through outlier detection using linear programming.


The address information is for the authors that have a + after their name.
Authors who are presenting talks have a * after their name.

Back to the full JSM 2010 program




2010 JSM Online Program Home

For information, contact jsm@amstat.org or phone (888) 231-3473.

If you have questions about the Continuing Education program, please contact the Education Department.