Neyman-pearson classiffication under high-dimensional settings
Most existing binary classification methods target on the optimization of the overall classification risk and may fail to serve some real-world applications such as cancer diagnosis, where users are more concerned with the risk of misclassifying one specific class than the other. Neyman-Pearson (NP)...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Published: |
JMLR, Inc.
2018
|
| Online Access: | http://hdl.handle.net/1721.1/116907 https://orcid.org/0000-0003-3582-8898 |
| _version_ | 1811090778345177088 |
|---|---|
| author | Zhao, Anqi Feng, Yang Wang, Lie Tong, Xin |
| author2 | Massachusetts Institute of Technology. Department of Mathematics |
| author_facet | Massachusetts Institute of Technology. Department of Mathematics Zhao, Anqi Feng, Yang Wang, Lie Tong, Xin |
| author_sort | Zhao, Anqi |
| collection | MIT |
| description | Most existing binary classification methods target on the optimization of the overall classification risk and may fail to serve some real-world applications such as cancer diagnosis, where users are more concerned with the risk of misclassifying one specific class than the other. Neyman-Pearson (NP) paradigm was introduced in this context as a novel statistical framework for handling asymmetric type I/II error priorities. It seeks classifiers with a minimal type II error and a constrained type I error under a user specified level. This article is the first attempt to construct classifiers with guaranteed theoretical performance under the NP paradigm in high-dimensional settings. Based on the fundamental Neyman-Pearson Lemma, we used a plug-in approach to construct NP-Type classifiers for Naive Bayes models. The proposed classifiers satisfy the NP oracle inequalities, which are natural NP paradigm counterparts of the oracle inequalities in classical binary classification. Besides their desirable theoretical properties, we also demonstrated their numerical advantages in prioritized error control via both simulation and real data studies. |
| first_indexed | 2024-09-23T14:51:49Z |
| format | Article |
| id | mit-1721.1/116907 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T14:51:49Z |
| publishDate | 2018 |
| publisher | JMLR, Inc. |
| record_format | dspace |
| spelling | mit-1721.1/1169072022-10-01T23:01:10Z Neyman-pearson classiffication under high-dimensional settings Zhao, Anqi Feng, Yang Wang, Lie Tong, Xin Massachusetts Institute of Technology. Department of Mathematics Wang, Lie Most existing binary classification methods target on the optimization of the overall classification risk and may fail to serve some real-world applications such as cancer diagnosis, where users are more concerned with the risk of misclassifying one specific class than the other. Neyman-Pearson (NP) paradigm was introduced in this context as a novel statistical framework for handling asymmetric type I/II error priorities. It seeks classifiers with a minimal type II error and a constrained type I error under a user specified level. This article is the first attempt to construct classifiers with guaranteed theoretical performance under the NP paradigm in high-dimensional settings. Based on the fundamental Neyman-Pearson Lemma, we used a plug-in approach to construct NP-Type classifiers for Naive Bayes models. The proposed classifiers satisfy the NP oracle inequalities, which are natural NP paradigm counterparts of the oracle inequalities in classical binary classification. Besides their desirable theoretical properties, we also demonstrated their numerical advantages in prioritized error control via both simulation and real data studies. 2018-07-11T17:36:56Z 2018-07-11T17:36:56Z 2016-01 2018-02-23T16:03:09Z Article http://purl.org/eprint/type/JournalArticle 1532-4435 1533-7928 http://hdl.handle.net/1721.1/116907 Zhao, Anqi et al. "Neyman-Pearson Classification under High-Dimensional Settings." Journal of Machine Learning Research, 17, 2016, pp. 7469-7507 https://orcid.org/0000-0003-3582-8898 https://dl.acm.org/citation.cfm?id=3053494 Journal of Machine Learning Research Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf JMLR, Inc. Journal of Machine Learning Research |
| spellingShingle | Zhao, Anqi Feng, Yang Wang, Lie Tong, Xin Neyman-pearson classiffication under high-dimensional settings |
| title | Neyman-pearson classiffication under high-dimensional settings |
| title_full | Neyman-pearson classiffication under high-dimensional settings |
| title_fullStr | Neyman-pearson classiffication under high-dimensional settings |
| title_full_unstemmed | Neyman-pearson classiffication under high-dimensional settings |
| title_short | Neyman-pearson classiffication under high-dimensional settings |
| title_sort | neyman pearson classiffication under high dimensional settings |
| url | http://hdl.handle.net/1721.1/116907 https://orcid.org/0000-0003-3582-8898 |
| work_keys_str_mv | AT zhaoanqi neymanpearsonclassifficationunderhighdimensionalsettings AT fengyang neymanpearsonclassifficationunderhighdimensionalsettings AT wanglie neymanpearsonclassifficationunderhighdimensionalsettings AT tongxin neymanpearsonclassifficationunderhighdimensionalsettings |