Causal-StoNet: Causal Inference for High-Dimensional Complex Data.

Area of Science:

• Data Science
• Machine Learning
• Causal Inference

Background:

• High-dimensional and complex datasets are common.
• Existing causal inference methods struggle with high dimensionality and nonlinear data generation processes.

Purpose of the Study:

• To propose a novel causal inference approach for high-dimensional complex data.
• To address challenges posed by high dimensionality and unknown, nonlinear data generation processes.

Main Methods:

• Utilizes deep learning techniques, specifically sparse deep learning theory and stochastic neural networks.
• Coherently addresses high dimensionality and unknown data generation processes.
• Accommodates datasets with missing values.

Main Results:

• The proposed approach demonstrates superior performance compared to existing methods.
• Extensive numerical studies validate the effectiveness of the new method.

Conclusions:

• The novel deep learning-based approach offers a robust solution for causal inference in complex, high-dimensional data.
• This method advances causal inference capabilities in fields like medicine, econometrics, and social science.