2021
DOI: 10.1007/978-3-030-88013-2_43
|View full text |Cite
|
Sign up to set email alerts
|

CETransformer: Casual Effect Estimation via Transformer Based Representation Learning

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
20
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
3
2
2

Relationship

2
5

Authors

Journals

citations
Cited by 10 publications
(20 citation statements)
references
References 17 publications
0
20
0
Order By: Relevance
“…In many works, representation distribution balancing is combined with other domain ideas. An ABCEI [122] combines a GAN [112] with a mutual information estimator regularization structure to balance the covariate distributions of the treatment and control groups in the representation space; CETransformer [127] creates a balanced covariate representation using the attention mechanism; As TransTEE [76] extends the representation distribution balance method to Continuous, Structured, and Dose-Related treatments, it makes causal effect estimation a more open-ended problem; SCI [75] introduces the concept of a subspace as shown in Figure 4, integrating the covariates into a common subspace, a treatment subspace, and a control subspace simultaneously, thereby obtaining a balanced representation and two specific representations. Afterwards, the public representation is connected to the specific representation of the treatment group and of the control group, and two potential results are obtained from the reconstruction and prediction network.…”
Section: Representation Of Distribution Balance Methodsmentioning
confidence: 99%
See 2 more Smart Citations
“…In many works, representation distribution balancing is combined with other domain ideas. An ABCEI [122] combines a GAN [112] with a mutual information estimator regularization structure to balance the covariate distributions of the treatment and control groups in the representation space; CETransformer [127] creates a balanced covariate representation using the attention mechanism; As TransTEE [76] extends the representation distribution balance method to Continuous, Structured, and Dose-Related treatments, it makes causal effect estimation a more open-ended problem; SCI [75] introduces the concept of a subspace as shown in Figure 4, integrating the covariates into a common subspace, a treatment subspace, and a control subspace simultaneously, thereby obtaining a balanced representation and two specific representations. Afterwards, the public representation is connected to the specific representation of the treatment group and of the control group, and two potential results are obtained from the reconstruction and prediction network.…”
Section: Representation Of Distribution Balance Methodsmentioning
confidence: 99%
“…The VCNet [58] model, implements an estimator of continuous mean dose-response curves. As of May 2021, NCoRE [126] uses cross-treatment interaction modeling to understand the underlying causal processes that produce multiple treatment combinations.After that, CETransformer [127] uses Transformer [128] to characterize covariates, and the attention mechanism is focused on the correlation among covariates. Following that, DONUT [129] and DeR-CFR [70] optimize based on previous work.…”
Section: Overviewmentioning
confidence: 99%
See 1 more Smart Citation
“…Transformer-based architectures using attention operations [68] were also applied to solving the CATE estimating problem [29,30,31,32]. Ideas of applying the transfer learning technique to the CATE estimation were considered in [69,70,71,8].…”
Section: Related Workmentioning
confidence: 99%
“…This is only a small part of all publications which are devoted to solving the problem of estimating CATE. Various approaches were used for solving the problem, including the support vector machine [25], tree-based models [9], neural networks [26,27,8,28], transformers [29,30,31,32].…”
Section: Introductionmentioning
confidence: 99%