samsi-causalml
Causal Inference Working Group I: Machine Learning Methods for Causal Inference
This group will study the methods in the interface between machine learning and causal inference.
Potential topics
Inspired by NeurIPS 2019 Workshop topic list.
- Predicting counterfactual outcomes (e.g., for individual treatment effect estimation)
- Reinforcement Learning and Causal Inference (e.g., for dynamic treatment regimes)
- Causal transfer learning
- Estimation of (conditional) average treatment effects
- Policy learning
- De-biasing observational data
- Causal discovery
- Applications on text data (e.g., counterfactual story generation)
- Applications in medicine (e.g., personalized treatment, clinical trials, fMRI)
- Applications in social sciences and policy evaluation
Potential research ideas
- Neural machine translation (NMT) approach to time-series counterfactual prediction
- Double-robust learning as a multitask learning problem
- Counterfactual prediction on observational panel data with continuous-valued treatment
- Generated data augmentation approach to de-confounding