/iap-cidl

Causal Inference & Deep Learning, MIT IAP 2018

iap-cidl

Causal Inference & Deep Learning, MIT IAP 2018

Taught by Fredrik Johansson and Max Shen. Organized by Max Shen.

Class Schedule

1. Tuesday, January 16th: 5pm-6:30pm at Room 4-231

  • Causal Models and Statistical Models (MS)
  • Structural Causal Models and Interventional Distributions (MS)
  • Potential Outcomes Framework (FJ)
  • Counterfactual Inference (FJ)
  • Causal Effects (FJ)
  • Conditional Treatment Effects (FJ)
  • Distributional Shift (FJ)
  • Domain Adaptation (FJ)
  • Importance Sampling (FJ)
  • Model Misspecification (FJ)

2. Wednesday, January 17th: 5pm-6:30pm at Room 4-231

  • Counterfactual Inference, continued (FJ)
  • Potential Outcomes and Deep Style Transfer (MS)
  • Cause-Effect Discovery with... (MS)
    • Additive Noise Models and the Hilbert-Schmidt Independence Criterion
    • Convolutional Neural Nets
    • Conditional GANs
    • Randomized Causation Coefficient
    • Proxy Variables

3. Thursday, January 18th: 5pm-6:30pm at Room 4-231

  • Causal Aspects of Reinforcement Learning (FJ)
    • Policy Optimization (FJ)
    • Off-Policy Evaluation (FJ)
    • Batch Reinforcement Learning (FJ)

Material and Notes

Will be uploaded at a later time.

Primary References

  • Johansson, F. D., Shalit, U., & Sontag, D. (2016). Learning Representations for Counterfactual Inference. http://arxiv.org/abs/1605.03661

  • Peters, J. (2017). Elements of Causal Inference (Draft).

  • Lopez-paz, D., & Sch, B. (2015). Towards a Learning Theory of Cause-Effect Inference. Proceedings of the 32nd International Conference on Machine Learning.

  • Incomplete, will be expanded at a later time.