IGNITE: A Minimax Game Toward Learning Individual Treatment Effects from Networked Observational Data

Guo, Ruocheng; Li, Jundong; Li, Yichuan; Candan, K. Selçuk; Raglin, Adrienne; Liu, Huan

doi:10.24963/ijcai.2020/625

“…A shared loss between the networks force Φ h and Φ π to produce similar representations of the nodes/covariates. Chu et al (2021) and Guo et al (2020) improve on this approach by dealing with the technical challenges of learning representations that are both faithful to the graph and can estimate causal effects. Veitch et al (2019b) take a formal semiparametric approach to identify a causal effect on graphs.…”

Section: Conditioning On Latent Confounding Encoded In Networkmentioning

confidence: 99%

Deep Learning of Potential Outcomes

Koch¹,

Sainburg²,

Geraldo³

et al. 2021

Preprint

View full text Add to dashboard Cite

This review systematizes the emerging literature for causal inference using deep neural networks under the potential outcomes framework. It provides an intuitive introduction on how deep learning can be used to estimate/predict heterogeneous treatment effects and extend causal inference to settings where confounding is non-linear, time varying, or encoded in text, networks, and images. To maximize accessibility, we also introduce prerequisite concepts from causal inference and deep learning. The survey differs from other treatments of deep learning and causal inference in its sharp focus on observational causal estimation, its extended exposition of key algorithms, and its detailed tutorials for implementing, training, and selecting among deep estimators in Tensorflow 2 available at github.com/kochbj/Deep-Learning-for-Causal-Inference.

show abstract

“…The network deconfounder [6] learns representations of confounders from network data by adopting the graph convolutional networks. Another work utilizes graph attention networks to learn representations and mitigates confounding bias by representation balancing and treatment prediction, simultaneously [5]. Causal network embedding (CNE) [20] is proposed to learn node embeddings from network data to represent confounders by reducing the causal estimation problem to a semi-supervised prediction of both the treatments and outcomes.…”

Section: Related Workmentioning

confidence: 99%

Graph Infomax Adversarial Learning for Treatment Effect Estimation with Networked Observational Data

Chu,

Rathbun,

Li

2021

Preprint

View full text Add to dashboard Cite

Treatment effect estimation from observational data is a critical research topic across many domains. The foremost challenge in treatment effect estimation is how to capture hidden confounders. Recently, the growing availability of networked observational data offers a new opportunity to deal with the issue of hidden confounders. Unlike networked data in traditional graph learning tasks, such as node classification and link detection, the networked data under the causal inference problem has its particularity, i.e., imbalanced network structure. In this paper, we propose a Graph Infomax Adversarial Learning (GIAL) model for treatment effect estimation, which makes full use of the network structure to capture more information by recognizing the imbalance in network structure. We evaluate the performance of our GIAL model on two benchmark datasets, and the results demonstrate superiority over the state-of-the-art methods.

show abstract

“…A shared loss between the networks force Φ h and Φ π to produce similar representations of the nodes/covariates. Chu et al (2021) and Guo et al (2020) improve on this approach by dealing with the technical challenges of learning representations that are both faithful to the graph and can estimate causal effects. Veitch et al (2019b) take a formal semiparametric approach to identify a causal effect on graphs.…”

Section: Conditioning On Latent Confounding Encoded In Networkmentioning

confidence: 99%

Deep Learning for Causal Inference

Koch¹,

Sainburg²,

Geraldo³

et al. 2021

Preprint

2

0

View full text Add to dashboard Cite

This review systematizes the emerging literature for causal inference using deep neural networks under the potential outcomes framework. It provides an intuitive introduction on how deep learning can be used to estimate/predict heterogeneous treatment effects and extend causal inference to settings where confounding is non-linear, time varying, or encoded in text, networks, and images. To maximize accessibility, we also introduce prerequisite concepts from causal inference and deep learning. The survey differs from other treatments of deep learning and causal inference in its sharp focus on observational causal estimation, its extended exposition of key algorithms, and its detailed tutorials for implementing, training, and selecting among deep estimators in Tensorflow 2 available at github.com/kochbj/Deep-Learning-for-Causal-Inference.

show abstract

IGNITE: A Minimax Game Toward Learning Individual Treatment Effects from Networked Observational Data

Cited by 13 publications

References 11 publications

Deep Learning of Potential Outcomes

Deep Learning of Potential Outcomes

Graph Infomax Adversarial Learning for Treatment Effect Estimation with Networked Observational Data

Deep Learning for Causal Inference

Contact Info

Product

Resources

About