The optimal crowd learning machine
Abstract Background Any family of learning machines can be combined into a single learning machine using various methods with myriad degrees of usefulness. Results For making predictions on an outcome, it is provably at least as good as the best machine in the family, given sufficient data. And if o...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
BMC
2017-05-01
|
Series: | BioData Mining |
Subjects: | |
Online Access: | http://link.springer.com/article/10.1186/s13040-017-0135-7 |
_version_ | 1811266759331676160 |
---|---|
author | Bilguunzaya Battogtokh Majid Mojirsheibani James Malley |
author_facet | Bilguunzaya Battogtokh Majid Mojirsheibani James Malley |
author_sort | Bilguunzaya Battogtokh |
collection | DOAJ |
description | Abstract Background Any family of learning machines can be combined into a single learning machine using various methods with myriad degrees of usefulness. Results For making predictions on an outcome, it is provably at least as good as the best machine in the family, given sufficient data. And if one machine in the family minimizes the probability of misclassification, in the limit of large data, then Optimal Crowd does also. That is, the Optimal Crowd is asymptotically Bayes optimal if any machine in the crowd is such. Conclusions The only assumption needed for proving optimality is that the outcome variable is bounded. The scheme is illustrated using real-world data from the UCI machine learning site, and possible extensions are proposed. |
first_indexed | 2024-04-12T20:49:06Z |
format | Article |
id | doaj.art-5cbcdc6f0e8e4ce0877d2c08889d9158 |
institution | Directory Open Access Journal |
issn | 1756-0381 |
language | English |
last_indexed | 2024-04-12T20:49:06Z |
publishDate | 2017-05-01 |
publisher | BMC |
record_format | Article |
series | BioData Mining |
spelling | doaj.art-5cbcdc6f0e8e4ce0877d2c08889d91582022-12-22T03:17:11ZengBMCBioData Mining1756-03812017-05-0110111210.1186/s13040-017-0135-7The optimal crowd learning machineBilguunzaya Battogtokh0Majid Mojirsheibani1James Malley2Center for Information Technology, National Institutes of HealthDepartment of Mathematics, California State University NorthridgeCenter for Information Technology, National Institutes of HealthAbstract Background Any family of learning machines can be combined into a single learning machine using various methods with myriad degrees of usefulness. Results For making predictions on an outcome, it is provably at least as good as the best machine in the family, given sufficient data. And if one machine in the family minimizes the probability of misclassification, in the limit of large data, then Optimal Crowd does also. That is, the Optimal Crowd is asymptotically Bayes optimal if any machine in the crowd is such. Conclusions The only assumption needed for proving optimality is that the outcome variable is bounded. The scheme is illustrated using real-world data from the UCI machine learning site, and possible extensions are proposed.http://link.springer.com/article/10.1186/s13040-017-0135-7Support Vector MachineTraining DataRandom ForestTest PointIndividual Machine |
spellingShingle | Bilguunzaya Battogtokh Majid Mojirsheibani James Malley The optimal crowd learning machine BioData Mining Support Vector Machine Training Data Random Forest Test Point Individual Machine |
title | The optimal crowd learning machine |
title_full | The optimal crowd learning machine |
title_fullStr | The optimal crowd learning machine |
title_full_unstemmed | The optimal crowd learning machine |
title_short | The optimal crowd learning machine |
title_sort | optimal crowd learning machine |
topic | Support Vector Machine Training Data Random Forest Test Point Individual Machine |
url | http://link.springer.com/article/10.1186/s13040-017-0135-7 |
work_keys_str_mv | AT bilguunzayabattogtokh theoptimalcrowdlearningmachine AT majidmojirsheibani theoptimalcrowdlearningmachine AT jamesmalley theoptimalcrowdlearningmachine AT bilguunzayabattogtokh optimalcrowdlearningmachine AT majidmojirsheibani optimalcrowdlearningmachine AT jamesmalley optimalcrowdlearningmachine |