Counterfactual Analysis

Counterfactual analysis explores "what if" scenarios by investigating how outcomes would change under hypothetical interventions or alternative choices. Current research focuses on developing methods to generate plausible counterfactuals, particularly within machine learning models (e.g., using structural causal models, diffusion models, and normalizing flows) and applying these techniques to diverse fields like reinforcement learning, bias detection in AI, and causal inference in healthcare and economics. This work aims to improve model explainability, identify and mitigate biases, and enable more robust and reliable decision-making across various applications.