This is a place to keep track of some random projects that would be interesting to work on. These are not my main priorities, but it feels nice to have a backlog somewhere of stuff to do on a rainy day.
vague not especially promising alignment directions
- Formalize “if statements / low pr” defection problems.
- Separate out “goodness is simple” and “performance is complex”.
- Delegation of computation, e.g., this ?
Theoretical ML
- Read some of Murphy’s book. e.g.,
- PAC-Bayes
- Generalization, 17 + 17.4
Empirical ML
- Finish ARENA
- What are transcoders?
TCS Research
IQPs (collab with N)
Misc fun TCS Problems
- PDSG
- P2P problem
- Triangles
- memory reallocation
- hashing
- Can you make a PRF with the property that for any two distinct keys,
are decently far apart? - Madeline’s questions
- cuckoo hashing
- neekon backdoors?
- catalytic stuff?