机器学习中的稳健数据采样:训练和验证数据选择的博弈论框架

IF 0.6 Q4 ECONOMICS
Games Pub Date : 2023-01-30 DOI:10.3390/g14010013
Zhaobin Mo, Xuan Di, Rongye Shi
{"title":"机器学习中的稳健数据采样:训练和验证数据选择的博弈论框架","authors":"Zhaobin Mo, Xuan Di, Rongye Shi","doi":"10.3390/g14010013","DOIUrl":null,"url":null,"abstract":"How to sample training/validation data is an important question for machine learning models, especially when the dataset is heterogeneous and skewed. In this paper, we propose a data sampling method that robustly selects training/validation data. We formulate the training/validation data sampling process as a two-player game: a trainer aims to sample training data so as to minimize the test error, while a validator adversarially samples validation data that can increase the test error. Robust sampling is achieved at the game equilibrium. To accelerate the searching process, we adopt reinforcement learning aided Monte Carlo trees search (MCTS). We apply our method to a car-following modeling problem, a complicated scenario with heterogeneous and random human driving behavior. Real-world data, the Next Generation SIMulation (NGSIM), is used to validate this method, and experiment results demonstrate the sampling robustness and thereby the model out-of-sample performance.","PeriodicalId":35065,"journal":{"name":"Games","volume":"14 1","pages":"13"},"PeriodicalIF":0.6000,"publicationDate":"2023-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"Robust Data Sampling in Machine Learning: A Game-Theoretic Framework for Training and Validation Data Selection\",\"authors\":\"Zhaobin Mo, Xuan Di, Rongye Shi\",\"doi\":\"10.3390/g14010013\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"How to sample training/validation data is an important question for machine learning models, especially when the dataset is heterogeneous and skewed. In this paper, we propose a data sampling method that robustly selects training/validation data. We formulate the training/validation data sampling process as a two-player game: a trainer aims to sample training data so as to minimize the test error, while a validator adversarially samples validation data that can increase the test error. Robust sampling is achieved at the game equilibrium. To accelerate the searching process, we adopt reinforcement learning aided Monte Carlo trees search (MCTS). We apply our method to a car-following modeling problem, a complicated scenario with heterogeneous and random human driving behavior. Real-world data, the Next Generation SIMulation (NGSIM), is used to validate this method, and experiment results demonstrate the sampling robustness and thereby the model out-of-sample performance.\",\"PeriodicalId\":35065,\"journal\":{\"name\":\"Games\",\"volume\":\"14 1\",\"pages\":\"13\"},\"PeriodicalIF\":0.6000,\"publicationDate\":\"2023-01-30\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Games\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3390/g14010013\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ECONOMICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Games","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3390/g14010013","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ECONOMICS","Score":null,"Total":0}
引用次数: 1

摘要

如何对训练/验证数据进行采样是机器学习模型的一个重要问题,特别是当数据集是异构和倾斜的时候。在本文中,我们提出了一种稳健地选择训练/验证数据的数据采样方法。我们将训练/验证数据的采样过程描述为一个双人游戏:训练器的目标是对训练数据进行采样,以最小化测试误差,而验证器的目标是对验证数据进行逆向采样,从而增加测试误差。在博弈平衡点上实现了鲁棒抽样。为了加速搜索过程,我们采用了强化学习辅助蒙特卡罗树搜索(MCTS)。我们将我们的方法应用于汽车跟随建模问题,这是一个复杂的场景,具有异质和随机的人类驾驶行为。利用下一代仿真(NGSIM)的实际数据验证了该方法,实验结果证明了该方法的采样鲁棒性,从而证明了模型的样本外性能。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Robust Data Sampling in Machine Learning: A Game-Theoretic Framework for Training and Validation Data Selection
How to sample training/validation data is an important question for machine learning models, especially when the dataset is heterogeneous and skewed. In this paper, we propose a data sampling method that robustly selects training/validation data. We formulate the training/validation data sampling process as a two-player game: a trainer aims to sample training data so as to minimize the test error, while a validator adversarially samples validation data that can increase the test error. Robust sampling is achieved at the game equilibrium. To accelerate the searching process, we adopt reinforcement learning aided Monte Carlo trees search (MCTS). We apply our method to a car-following modeling problem, a complicated scenario with heterogeneous and random human driving behavior. Real-world data, the Next Generation SIMulation (NGSIM), is used to validate this method, and experiment results demonstrate the sampling robustness and thereby the model out-of-sample performance.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Games
Games Decision Sciences-Statistics, Probability and Uncertainty
CiteScore
1.60
自引率
11.10%
发文量
65
审稿时长
11 weeks
期刊介绍: Games (ISSN 2073-4336) is an international, peer-reviewed, quick-refereeing open access journal (free for readers), which provides an advanced forum for studies related to strategic interaction, game theory and its applications, and decision making. The aim is to provide an interdisciplinary forum for all behavioral sciences and related fields, including economics, psychology, political science, mathematics, computer science, and biology (including animal behavior). To guarantee a rapid refereeing and editorial process, Games follows standard publication practices in the natural sciences.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信