To solve Rubik's cube in generality, you need to approach it as a reinforcement learning problem. You must explore the search space somehow, e.g., MCTS, Thompson sampling, etc. Is all search in discrete spaces "symbolic"? Not in an interesting way
Many people have tried to defend pure DRL w Nature Machine Intelligence article that actually is a hybrid model; the below youtube pointer IS to a pure convnet - but read fine print: “network works decently well for any position less than 6 moves away from solved”
#symbolphobiahttps://twitter.com/AlexRoseGames/status/1186571935611850752 …
-
-
-
Not saying it is impossible, but I am aware of no pure DRL solution to Rubik's thus far; everything I have seen is either a hybrid or less reliable than a pure symbolic solution.
- 1 more reply
New conversation -
-
-
btw, this one solves visual puzzles without innate knowledge (31 step instance in 8puzzle, 15 step hanoi...) by discrete repr learning + Dijkstra. I should try Rubik too... https://www.aaai.org/ocs/index.php/AAAI/AAAI18/paper/download/16302/16179 …
- 1 more reply
New conversation -
-
-
Interesting number. Reminds me of some of Richard Feynman’s work. Does is move into the same math as fluid dynamics? I’m just now jumping into the DRL pool, but I like where it can go. The math keeps leading to Torricelli’s trumpet to balance. Our brains fool us to keep us sane.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.