P
Initializing...
The information budget of a k-valued optimal action is at most $\log k$ · Prove2Me