Sciweavers

Free Online Productivity Tools i2Speak i2Symbol i2OCR iTex2Img iWeb2Print iWeb2Shot i2Type iPdf2Split iPdf2Merge i2Bopomofo i2Arabic i2Style i2Image i2PDF iLatex2Rtf Sci2ools

10

ICML
2000
IEEE

favoriteEmaildiscussreport

151views Machine Learning» more ICML 2000»

Bayesian Averaging of Classifiers and the Overfitting Problem

14 years 5 months ago

Bayesian Averaging of Classifiers and the Overfitting Problem

Download www.cs.washington.edu

Although Bayesian model averaging is theoretically the optimal method for combining learned models, it has seen very little use in machine learning. In this paper we study its application to combining rule sets, and compare it with bagging and partitioning, two popular but more ad hoc alternatives. Our experiments show that, surprisingly, Bayesian model averaging's error rates are consistently higher than the other methods'. Further investigation shows this to be due to a marked tendency to overfit on the part of Bayesian model averaging, contradicting previous beliefs that it solves (or avoids) the overfitting problem.

Pedro Domingos

Real-time Traffic

Ad Hoc Alternatives | Bayesian Model Averaging | Further Investigation | ICML 2000 | Machine Learning |

claim paper

Related Content

» Learning Bayesian Belief Network Classifiers Algorithms and System

» Learning Multicategory Classification in Bayesian Framework

» Lazy Learning of Bayesian Rules

» Diagram Structure Recognition by Bayesian Conditional Random Fields

» Evolutionary Learning of Dynamic Naive Bayesian Classifiers

» Discriminative versus generative parameter and structure learning of Bayesian network clas...

» Probabilistic Brain Atlas Encoding Using Bayesian Inference

» Lazy Bayesian Rules A Lazy SemiNaive Bayesian Learning Technique Competitive to Boosting D...

» Combining Heterogeneous Classifiers for Network Intrusion Detection

Post Info
More Details (n/a)

Added	17 Nov 2009
Updated	17 Nov 2009
Type	Conference
Year	2000
Where	ICML
Authors	Pedro Domingos

Comments (0)