國立高雄大學圖資館 |

登入

回首頁

Topics in Causal Inference and Distribution Shifts /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Topics in Causal Inference and Distribution Shifts /Yujin Jeong.
作者:	Jeong, Yujin,
面頁冊數:	1 electronic resource (169 pages)
附註:	Source: Dissertations Abstracts International, Volume: 87-04, Section: A.
附註:	Advisors: Fox, Emily Committee members: Johari, Ramesh; Rothenhäusler, Dominik; Bent, Stacey F.
Contained By:	Dissertations Abstracts International87-04A.
標題:	Gene expression.
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32149631
ISBN:	9798290649177

Topics in Causal Inference and Distribution Shifts /
Jeong, Yujin,

Topics in Causal Inference and Distribution Shifts /Yujin Jeong. - 1 electronic resource (169 pages)

Source: Dissertations Abstracts International, Volume: 87-04, Section: A.

Randomized experiments represent the gold standard for causal inference, allowing scientists to estimate treatment effects by comparing outcomes between randomly assigned treatment and control groups. While randomized experiments have traditionally focused on low-dimensional outcomes, technological advances have expanded the scope and complexity of measured outcomes.With the development of sensing technologies, randomized trials with primary outcomes represented as high-dimensional vectors have become increasingly prevalent. For example, these outcomes could be week-long time-series data from wearable devices or high-dimensional neuroimaging data, such as from functional magnetic resonance imaging. In Chapter 2, we focus on randomized experiments with such high-dimensional outcomes characterized by sparse treatment effects, where interventions may influence only a small number of dimensions (e.g., specific temporal windows or targeted brain regions).Conventional practices, such as using fixed, low-dimensional summaries of the outcomes, often result in significantly reduced power for detecting treatment effects. To address this limitation, we consider a large collection of outcome summaries, potentially with dimensionality exceeding the sample size. However, with such high-dimensional outcome summaries, a standard multiple testing approach may lead to a decrease in power due to the need for large correction factors.Instead, we propose a procedure that involves subset selection followed by inference. Given a large set of outcome summaries, we identify the subset that captures the treatment effects-requiring only a single call to the Lasso-and then conduct inference on the selected subset. We theoretically demonstrate that our method consistently recovers the correct subset, and empirically show that it can improve power by first identifying the relevant subset and focusing inference on the selected dimensions. This chapter is based on joint work with Emily Fox and Ramesh Johari.

English

ISBN: 9798290649177Subjects--Topical Terms:

188888
Gene expression.
Subjects--Index Terms:

Subset selection

Topics in Causal Inference and Distribution Shifts /
LDR:03384nmm a22003973i 4500 001 693737
005 20260511063017.5
006 m o d
007 cr|nu||||||||
008 260819s2025 miu||||||m |||||||eng d
020 $a 9798290649177
035 $a (MiAaPQD)AAI32149631
035 $a (MiAaPQD)Stanfordcr335jv9786
035 $a AAI32149631
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Jeong, Yujin, $e author. $3 1010387
245 1 0 $a Topics in Causal Inference and Distribution Shifts / $c Yujin Jeong.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (169 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: Dissertations Abstracts International, Volume: 87-04, Section: A.
500 $a Advisors: Fox, Emily Committee members: Johari, Ramesh; Rothenhäusler, Dominik; Bent, Stacey F.
502 $b Ph.D. $c Stanford University $d 2025.
520 $a Randomized experiments represent the gold standard for causal inference, allowing scientists to estimate treatment effects by comparing outcomes between randomly assigned treatment and control groups. While randomized experiments have traditionally focused on low-dimensional outcomes, technological advances have expanded the scope and complexity of measured outcomes.With the development of sensing technologies, randomized trials with primary outcomes represented as high-dimensional vectors have become increasingly prevalent. For example, these outcomes could be week-long time-series data from wearable devices or high-dimensional neuroimaging data, such as from functional magnetic resonance imaging. In Chapter 2, we focus on randomized experiments with such high-dimensional outcomes characterized by sparse treatment effects, where interventions may influence only a small number of dimensions (e.g., specific temporal windows or targeted brain regions).Conventional practices, such as using fixed, low-dimensional summaries of the outcomes, often result in significantly reduced power for detecting treatment effects. To address this limitation, we consider a large collection of outcome summaries, potentially with dimensionality exceeding the sample size. However, with such high-dimensional outcome summaries, a standard multiple testing approach may lead to a decrease in power due to the need for large correction factors.Instead, we propose a procedure that involves subset selection followed by inference. Given a large set of outcome summaries, we identify the subset that captures the treatment effects-requiring only a single call to the Lasso-and then conduct inference on the selected subset. We theoretically demonstrate that our method consistently recovers the correct subset, and empirically show that it can improve power by first identifying the relevant subset and focusing inference on the selected dimensions. This chapter is based on joint work with Emily Fox and Ramesh Johari.
546 $a English
590 $a School code: 0212
650 4 $a Gene expression. $3 188888
650 4 $a Statistics. $3 182057
653 $a Subset selection
653 $a Causal inference
690 $a 0501
690 $a 0463
710 2 $a Stanford University. $3 212607
720 1 $a Fox, Emily $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 87-04A.
790 $a 0212
791 $a Ph.D.
792 $a 2025
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=32149631