A review of causal inference methods in computer science and their application in analyzing complex and multidimensional data.

Fatemeh Safaie¹ Zahra Safaie²

1) Bachelor of Computer Science, Azad University, Najafabad Branch
2) Bachelor of Computer Science, Azad University, Najafabad Branch

IOI Code : AADG-ABZCCA
Publication : The second international conference on new approaches in engineering and basic sciences(icnabs.ir/2nd)
Abstract :
    Causal inference plays an increasing role in computer science and complex data analysis as a framework for understanding cause-and-effect relationships from observational, experimental, or simulated data. While classical statistical methods focus mainly on correlations, causal inference seeks to uncover intervention effects and answer “why” and “what if” questions. This paper systematically reviews the theoretical foundations, methods, algorithms, and applications of causal inference in various computer science domains such as machine learning, data mining, computer vision, natural language processing, and recommender systems. Given the characteristics of modern complex and multidimensional data (such as high dimensionality, heterogeneity, spatiotemporal dependencies, and graph structures), challenges such as high dimensionality, confounding, selection bias, and noise are addressed. Methods based on Causal Graphical Models, Counterfactual Reasoning, and Structural Causal Models are described. Also, the application of these methods to the analysis of multidimensional data such as images, videos, text data, and social networks is presented. Finally, future trends, including the integration of deep learning with causal inference, the scalability of algorithms, and ethics in causal modeling are discussed.

Keywords : causal inference causal graphical models causality in machine learning multidimensional data