Author Description

Causal Machine Learning (CausalML) is an umbrella term for machine learning methods that formalize the data generation process as a causal model. This perspective enables one to reason about the effects of changes to this process (interventions) and what would have happened in hindsight (counterfactuals). CausalML can be categorized into five groups according to the problems they address, namely (1) causal supervised learning, (2) causal generative modeling, (3) causal explanations, (4) causal fairness, and (5) causal reinforcement learning. In this monograph, approaches in the five categories of CausalML are systematically compared, and open problems are identified. The field-specific applications in computer vision, natural language processing, and graph representation learning are reviewed. Further, an overview of causal benchmarks is provided, as well as a discussion of the state of this nascent field, including recommendations for future work.