Data Mining Techniequs that aware social fairness

2014 Fiscal Year Final Research Report

• Project/Area Number: 24500194
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: National Institute of Advanced Industrial Science and Technology
• Principal Investigator: KAMISHIMA Toshihiro 独立行政法人産業技術総合研究所, ヒューマンライフテクノロジー研究部門, 主任研究員 (50356820)
• Co-Investigator(Kenkyū-buntansha): AKAHO Shotaro 独立行政法人産業技術総合研究所, ヒューマンライフテクノロジー研究部門, 研究グループ長 (40356340)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: データマイニング / 機械学習 / 統計 / 独立性 / 公正性

Outline of Final Research Achievements

As the improvement of data mining technologies, such technologies are being used for determinations that seriously affect individual' lives, such as employment applications, credit, insurance rates, and so on. Needless to say, such serious determinations must be unbiased and fair in terms of socially sensitive information, such as gender or race. For this purpose, we develop the technique to avoid the use of such sensitive information in the serious determinations. The simple elimination of sensitive information from the calculations is insufficient for avoiding inappropriate determination process, due to the indirect influence of sensitive information. This phenomena is called the read-lining effect. We treat this red-lining effect by introducing the methodologies based on the notion of the statistical independence.

Free Research Field

データマイニング