Given that the material I have planned goes a bit faster than anticipated, the last two hours of the first module could be devoted to various things.
For the moment, I have planned to present you a learning/classification method (Naive credal classifier) using imprecise probabilities in a bit more details, possibly with a demo (that you could also do on your computer) based on https://github.com/sdestercke/classifip.
Other possibilities include:
- Some small exercice on what we have seen (modeling and solving a toy decision problem).
- A general talk about some on-going research within the field (do not expect something too structured in this case, even if you can hope for it).
If it is fine with you, I will proceed with my initial ideas, otherwise we can discuss it here or in the classroom next week.