Statistical Learning for Data Science
The process of learning models and inferring quantities from data lies at the heart of several fields of study, including machine learning, signal processing, control theory, and statistics. The goal of this PhD-level course is to provide a solid statistical foundation for researchers in related fields.
We will tackle certain important issues in parameter inference and policy evaluation that are not adequately addressed in conventional machine learning and statistics textbooks.
Course schedule
| Lecture | Date | Room | Suppl. reading |
|---|---|---|---|
| Fundamental concepts | Sept 9 @ 10 | Ång 100155 | F: 1.1-3, 1.6. W: 1-5. vdV: 5. |
| Inference and model learning | Sept 16 @ 10 | Ång 100155 | F: 5.8 |
| Validation, regularization, Bayesian approach | Sept 23 @ 10 | Ång 100155 | F: 1.8 |
| Recursive methods, adaptation, time series data | Sept 30 @ 10 | Ång 100155 | L: 3, 4, 7, 8 |
| Policies and causal structures | Oct 7 @ 10 | Ång 100155 | P: 6, W: 17 |
| Predictive policies | Oct 14 @ 10 | Ång 100155 | W: 22, F: 5.1 |
| Interventional policies, summary | Oct 21 @ 10 | Ång 101190 | P: 6 |
References:
- [F] T. S. Ferguson, Mathematical Statistics, Academic Press, 1967.
- [W] L. Wasserman, All of Statistics, Springer 2004.
- [P] J. Peters, et al, Elements of Causal Inference, MIT Press 2017.
- [vdV] A.W. van der Vaart, Asymptotic Statistics, CUP 1998.
- [L] L. Ljung, System Identification (2nd ed.), Pearson 1999.
Examination
- Weekly homeworks (7hp)
- (Optional project corresponding to a three-page research paper (+3hp))
Prerequisites
Undergraduate courses in linear algebra and probability theory.
Course material
Available here
Registration
E-mail Dave Zachariah.