國立虎尾科技大學 |

Measurement Non-Invariance in Machine Learning : = An Intersection of Machine Learning Bias and Test Bias.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Measurement Non-Invariance in Machine Learning :/
其他題名:	An Intersection of Machine Learning Bias and Test Bias.
作者:	Suzuki, Honoka.
面頁冊數:	1 online resource (115 pages)
附註:	Source: Masters Abstracts International, Volume: 85-07.
Contained By:	Masters Abstracts International85-07.
標題:	Psychology. -
電子資源:	click for full text (PQDT)
ISBN:	9798381384123

Measurement Non-Invariance in Machine Learning : = An Intersection of Machine Learning Bias and Test Bias.
Suzuki, Honoka.

Measurement Non-Invariance in Machine Learning :An Intersection of Machine Learning Bias and Test Bias. - 1 online resource (115 pages)

Source: Masters Abstracts International, Volume: 85-07.

Thesis (M.A.)--The University of North Carolina at Chapel Hill, 2023.

Includes bibliographical references

Algorithmic and machine learning bias have stirred concern in society as machine learning continues to channel into sensitive and high-stakes applications, including in healthcare, hiring, and criminal justice. While research surrounding machine learning bias may be relatively new, psychometricians have for decades researched a closely paralleled topic of test bias in psychological and educational testing. Leveraging the connection between these two fairness domains, this thesis studies the problem of machine learning bias from a measurement perspective, specifically focusing on measurement non-invariance in outcome variables as a source of machine learning bias. A framework is introduced, which conceptualizes machine learning bias in a psychometric sense and allows for tests of measurement invariance in machine learning. Using a Monte Carlo simulation study, the consequences of measurement bias on machine learning bias are demonstrated, as well as the effectiveness of a proposed bias mitigation technique to address these effects of measurement bias, which also follows from the proposed framework. The application of the proposed methods is illustrated with data from a large-scale health survey. Broader implications of the relevance of fairness in measurement for fairness in machine learning are discussed.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798381384123Subjects--Topical Terms:

555998
Psychology.
Subjects--Index Terms:

Algorithmic fairnessIndex Terms--Genre/Form:

554714
Electronic books.

Measurement Non-Invariance in Machine Learning : = An Intersection of Machine Learning Bias and Test Bias.
LDR:02755ntm a22004097 4500 001 1146633
005 20240812064701.5
006 m o d
007 cr bn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798381384123
035 $a (MiAaPQ)AAI30689446
035 $a AAI30689446
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Suzuki, Honoka. $3 1472058
245 1 0 $a Measurement Non-Invariance in Machine Learning : $b An Intersection of Machine Learning Bias and Test Bias.
264 0 $c 2023
300 $a 1 online resource (115 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Masters Abstracts International, Volume: 85-07.
500 $a Advisor: Gonzalez, Oscar.
502 $a Thesis (M.A.)--The University of North Carolina at Chapel Hill, 2023.
504 $a Includes bibliographical references
520 $a Algorithmic and machine learning bias have stirred concern in society as machine learning continues to channel into sensitive and high-stakes applications, including in healthcare, hiring, and criminal justice. While research surrounding machine learning bias may be relatively new, psychometricians have for decades researched a closely paralleled topic of test bias in psychological and educational testing. Leveraging the connection between these two fairness domains, this thesis studies the problem of machine learning bias from a measurement perspective, specifically focusing on measurement non-invariance in outcome variables as a source of machine learning bias. A framework is introduced, which conceptualizes machine learning bias in a psychometric sense and allows for tests of measurement invariance in machine learning. Using a Monte Carlo simulation study, the consequences of measurement bias on machine learning bias are demonstrated, as well as the effectiveness of a proposed bias mitigation technique to address these effects of measurement bias, which also follows from the proposed framework. The application of the proposed methods is illustrated with data from a large-scale health survey. Broader implications of the relevance of fairness in measurement for fairness in machine learning are discussed.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Psychology. $3 555998
650 4 $a Statistics. $3 556824
650 4 $a Educational tests & measurements. $3 1180442
650 4 $a Quantitative psychology. $3 1182802
653 $a Algorithmic fairness
653 $a Machine learning bias
653 $a Measurement invariance
653 $a Psychometrics
653 $a Test bias
655 7 $a Electronic books. $2 local $3 554714
690 $a 0632
690 $a 0621
690 $a 0288
690 $a 0463
710 2 $a The University of North Carolina at Chapel Hill. $b Psychology. $3 1468856
710 2 $a ProQuest Information and Learning Co. $3 1178819
773 0 $t Masters Abstracts International $g 85-07.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30689446 $z click for full text (PQDT)