Saved in:
Bibliographic Details
Main Authors: da Cunha, Arthur, Larsen, Kasper Green, Ritzert, Martin
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.02976
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913620111130624
author da Cunha, Arthur
Larsen, Kasper Green
Ritzert, Martin
author_facet da Cunha, Arthur
Larsen, Kasper Green
Ritzert, Martin
contents In boosting, we aim to leverage multiple weak learners to produce a strong learner. At the center of this paradigm lies the concept of building the strong learner as a voting classifier, which outputs a weighted majority vote of the weak learners. While many successful boosting algorithms, such as the iconic AdaBoost, produce voting classifiers, their theoretical performance has long remained sub-optimal: The best known bounds on the number of training examples necessary for a voting classifier to obtain a given accuracy has so far always contained at least two logarithmic factors above what is known to be achievable by general weak-to-strong learners. In this work, we break this barrier by proposing a randomized boosting algorithm that outputs voting classifiers whose generalization error contains a single logarithmic dependency on the sample size. We obtain this result by building a general framework that extends sample compression methods to support randomized learning algorithms based on sub-sampling.
format Preprint
id arxiv_https___arxiv_org_abs_2402_02976
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Boosting, Voting Classifiers and Randomized Sample Compression Schemes
da Cunha, Arthur
Larsen, Kasper Green
Ritzert, Martin
Machine Learning
In boosting, we aim to leverage multiple weak learners to produce a strong learner. At the center of this paradigm lies the concept of building the strong learner as a voting classifier, which outputs a weighted majority vote of the weak learners. While many successful boosting algorithms, such as the iconic AdaBoost, produce voting classifiers, their theoretical performance has long remained sub-optimal: The best known bounds on the number of training examples necessary for a voting classifier to obtain a given accuracy has so far always contained at least two logarithmic factors above what is known to be achievable by general weak-to-strong learners. In this work, we break this barrier by proposing a randomized boosting algorithm that outputs voting classifiers whose generalization error contains a single logarithmic dependency on the sample size. We obtain this result by building a general framework that extends sample compression methods to support randomized learning algorithms based on sub-sampling.
title Boosting, Voting Classifiers and Randomized Sample Compression Schemes
topic Machine Learning
url https://arxiv.org/abs/2402.02976