-
首页
-
- 首页
- WORKSHOP预告
数字金融workshop：Training Machine Learning to Anticipate Manipulation
发布日期：2024-05-28 12:00 来源：
时间/Time： 5月28日 周二 北京时间下午 14:00-15:30
Tuesday 22 May 2024, 2:00-3:30 p.m.
地点/Venue：
线上：北京大学国家发展研究院承泽园校区245教室
线下：ZOOM会议（会议号: 897 2560 1400 密码: 938759）
主讲人/Speaker： Joshua Blumenstock
主持人/Host： 胡佳胤 Jiayin Hu
Abstract
An increasing number of decisions are guided by machine learning algorithms. In many settings, from consumer credit to criminal justice, those decisions are made by applying an estimator to data on an individual's observed behavior. But when consequential decisions are encoded in rules, individuals may strategically alter their behavior to achieve desired outcomes. This paper develops an empirical approach to adjust decision algorithms to anticipate manipulation. By explicitly modeling incentives to manipulate, our approach produces decision rules that are stable under manipulation, even when the rules are fully transparent. We stress test this approach through a large field experiment in Kenya. When implemented, decision rules estimated with our strategy-robust approach outperform those based on standard machine learning approaches.
主讲人介绍/Biography:
Joshua Blumenstock is a Chancellor’s Associate Professor at the U.C. Berkeley School of Information and the Goldman School of Public Policy. He is a co-Director of the Global Policy Lab and the co-Director of the Center for Effective Global Action. Blumenstock does research at the intersection of machine learning and empirical economics, with a focus on how novel data and technology can better address the needs of poor and vulnerable people around the world. He has a Ph.D. in Information Science and a M.A. in Economics from U.C. Berkeley, and Bachelor’s degrees in Computer Science and Physics from Wesleyan University. He is a recipient of awards including the NSF CAREER award, the Intel Faculty Early Career Honor, and the U.C. Berkeley Chancellor's Award for Public Service. His work has appeared in general interest journals including Science, Nature, and Proceedings of the National Academy of Sciences, as well as top economics journals (e.g., the American Economic Review) and computer science conferences (e.g., ICML, KDD, AAAI, WWW, CHI).
国家发展研究院官方微信
Copyright© 1994-2012 北京大学 国家发展研究院 版权所有, 京ICP备05065075号-1
保留所有权利，不经允许请勿挪用
-
首页
-
数字金融workshop：Training Machine Learning to Anticipate Manipulation
发布日期：2024-05-28 12:00 来源：
时间/Time： 5月28日 周二 北京时间下午 14:00-15:30
Tuesday 22 May 2024, 2:00-3:30 p.m.
地点/Venue：
线上：北京大学国家发展研究院承泽园校区245教室
线下：ZOOM会议（会议号: 897 2560 1400 密码: 938759）
主讲人/Speaker： Joshua Blumenstock
主持人/Host： 胡佳胤 Jiayin Hu
Abstract
An increasing number of decisions are guided by machine learning algorithms. In many settings, from consumer credit to criminal justice, those decisions are made by applying an estimator to data on an individual's observed behavior. But when consequential decisions are encoded in rules, individuals may strategically alter their behavior to achieve desired outcomes. This paper develops an empirical approach to adjust decision algorithms to anticipate manipulation. By explicitly modeling incentives to manipulate, our approach produces decision rules that are stable under manipulation, even when the rules are fully transparent. We stress test this approach through a large field experiment in Kenya. When implemented, decision rules estimated with our strategy-robust approach outperform those based on standard machine learning approaches.
主讲人介绍/Biography:
Joshua Blumenstock is a Chancellor’s Associate Professor at the U.C. Berkeley School of Information and the Goldman School of Public Policy. He is a co-Director of the Global Policy Lab and the co-Director of the Center for Effective Global Action. Blumenstock does research at the intersection of machine learning and empirical economics, with a focus on how novel data and technology can better address the needs of poor and vulnerable people around the world. He has a Ph.D. in Information Science and a M.A. in Economics from U.C. Berkeley, and Bachelor’s degrees in Computer Science and Physics from Wesleyan University. He is a recipient of awards including the NSF CAREER award, the Intel Faculty Early Career Honor, and the U.C. Berkeley Chancellor's Award for Public Service. His work has appeared in general interest journals including Science, Nature, and Proceedings of the National Academy of Sciences, as well as top economics journals (e.g., the American Economic Review) and computer science conferences (e.g., ICML, KDD, AAAI, WWW, CHI).
访问学者申请
EMAIL
北京大学
ENGLISH
