conftrace_

2006 NeurIPS NeurIPS 2006

AdaBoost is Consistent

Abstract

The risk, or probability of error, of the classifier produced by the AdaBoost algorithm is investigated. In particular, we consider the stopping strategy to be used in AdaBoost to achieve universal consistency. We show that provided AdaBoost is stopped after n iterations--for sample size n and < 1--the sequence of risks of the classifiers it produces approaches the Bayes risk if Bayes risk L > 0.

🚀 Conference Pioneer - NeurIPS 2006

📈 Trend Setter - Ensemble Methods

🧭 Keyword Pioneer - bayes risk

🐣 Hot Topic Early Bird - ensemble learning

Authors

Peter L. Bartlett, Mikhail Traskin

Topics

Machine Learning > Core Methods > Classification Machine Learning > Optimization & Theory > Learning Theory Machine Learning > Optimization & Theory > Statistical Learning Machine Learning > Optimization & Theory > Theory Machine Learning > Core Methods > Ensemble Methods

Keywords

statistical learning learning theory statistical consistency ensemble learning adaboost consistency analysis risk minimization universal consistency boosting algorithm stopping criterion bayes risk boosting consistency adaboost algorithm

More from NeurIPS 2006

All NeurIPS 2006 papers →

Sparse Multinomial Logistic Regression via Bayesian L1 Regularisation

Gaussian and Wishart Hyperkernels

A Complexity-Distortion Approach to Joint Pattern Alignment

Cross-Validation Optimization for Large Scale Hierarchical Classification Kernel Methods

Effects of Stress and Genotype on Meta-parameter Dynamics in Reinforcement Learning

A Novel Gaussian Sum Smoother for Approximate Inference in Switching Linear Dynamical Systems