Something that "happened" recently in a livestreamed meeting:pic.twitter.com/MxtUvfekUH
-
-
i thought about this question at some point this year. the naive solution i think is to start with random rules over a 1D automata over pitch space (k=2 over 256 state cells for MIDI + 1 for endofsong) and do greedy rule assignment over new examples. when there's a rule collision
-
expand r by 1 over past states. but this can end up taking up much more space than the original encoding. it seems you can probably analytically bound your error tolerance and dramatically reduce the size of the ruleset, but i didn't do the work
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.