This is a linkpost for https://www.matsprogram.org/agent
Now that the applications are closed, is it possible to get some short overview of the intended ways to approach the application problems? For question 1 and question 3 I think I could cobble together some tedious proofs that don't carry much generalisable insight. For question 2 I so far don't have any method for arriving at an answer. I am curious about what kind of knowledge/skill the questions were verifying.
In MATS Winter 2023, I'm going to mentor scholars who want to work on the learning-theoretic agenda. This was not part of the original announcement, because (due to logistical reasons) I joined the programme in the last moment.
Apply here.
See here for more details about the broader programme.
Here are some examples of work by my scholars in the Winter 2022 cohort: 1 2 3. The most impressive example is not here because it's still unpublished. One of the scholars from that cohort now holds a long-term position working on the learning-theoretic agenda.
If you know someone else who might be suitable and interested to apply, please pass on the word!
Tangentially related: Although it was not officially announced, I had plans to create an internship programme for researchers interested to work on the learning-theoretic agenda, which would involve moving to Israel for a short period. This is now postponed indefinitely because of the war. If you heard of that plan and considered to apply, you might consider applying to my track in MATS instead.