Assignments
Four programming assignments, together worth 20% of the grade. Each is due at 6:00 pm on the date shown.
| Assignment | Topic | Marks | Due |
|---|---|---|---|
| Assignment 1 | Gradient descent with JAX; maximum likelihood for Normal, Student-t and exponential distributions | 5 | 10 Aug 2022 |
| Assignment 2 | MLE and MAP for linear and logistic regression; Normal, Laplace and Student-t choices | 6 | 23 Aug 2022 |
| Assignment 3 | Laplace approximation, Monte Carlo and probit approximation for logistic regression | 4 | 15 Sep 2022 |
| Assignment 4 | CNNs with MC dropout in Flax; Gaussian processes on the Mauna Loa CO2 data | 5 | 27 Oct 2022 |
Rules for every assignment
- Use TensorFlow Probability (TFP) with the JAX substrate for distributions.
- Use JAX for automatic differentiation and vectorised code.
- Work in groups of two.
- Submit a single Jupyter notebook.
- Draw insights from your results rather than only presenting answers. For example, if a question asks you to try a different likelihood, observe how it compares with the default likelihood.