Is there anywhere to see the games? I'm curious if the AI is superior to humans or just human trained AI. It'd also be interesting to see the source, but that is apparently not being released for some reason.
AlphaGo Zero: Learning from scratch
261–270 of 324 posts
Re: AlphaGo Zero: Learning from scratch
#262Is there anywhere to see the games? I'm curious if the AI is superior to humans or just human trained AI. It'd also be interesting to see the source, but that is apparently not being released for some reason.
It'd also be interesting to see the source, but that is apparently not being released for some reason.
Last time this was brought up, someone implied it's closed source so as not to boost the Chinese competitor.
Re: AlphaGo Zero: Learning from scratch
#263I remember reading ages ago in Scientific American about a much more interesting (and useful) AI application of this technique. Genetic algorithms were used to evolve new, more efficient variants of existing electronic circuits. I dug it up - it was: https://www.scientificamerican.com/magazine/sa/2003/02-01/#a... Article "Evolving inventions". I have no idea if there is an open-access version anywhere. As far as I re…
Nature is actually hosting it without a paywall:
https://www.nature.com/scientificamerican/journal/v288/n2/pd...
Re: AlphaGo Zero: Learning from scratch
#264Earlier quoted context omitted.
Until I see AlphaGo zero defeating StockFish 100-0 and with same algorithm defeating best Go AI and killing the Atari games including montezuma’s revenge, I call this hype bullshit. Give me your results on OpenAI gym in a variety of different styles of games including GTA and WoW. I will believe you if a generic unsupervised algorithm running on a single machine is absolutely destroying the best players. Until then .…
Just like Lee Se-dol is a Go grandmaster, beats Gary Kasparov at chess and can also get a perfect score in Pac-Man, right? I mean, if you can't do all of those things then are you even a human-level intelligence?
This just illustrates that surpassing "human level" performance is a silly and arbitrary benchmark, because there is no such thing as general human level performance. But I bet Kasparov would be pretty good at Go, and Sedol would be pretty good at chess.
Universality is the real hard problem of AI. In the long run, a mediocre AI that does a lot of different things is far more useful that most targeted "superhuman" AIs. Most domains simply don't require better-than-human performance, but could still reap tremendous benefits from automation.
Re: AlphaGo Zero: Learning from scratch
#265Earlier quoted context omitted.
> most of the problems required for a general AI aren't well-defined. Do you care to give an example? Are they more or less well defined than find-the-cat-in-the-picture problem? > Producing a definitive, objective score for a paper clip collection is very difficult if not impossible. Erm, producing of objective comparison of relative values of Go board positions is still not possible.
The absolute value of any Go board position is well-defined, and MCTS provides good computationally tractable approximations that get better as the rest of the system improves but already start better than random.
Check the Nature paper (and I think this is one of the biggest take-aways from AlphaGo Zero):
"Finally, it uses a simpler tree search that relies upon this single neural network to evaluate positions and sample moves, without performing any Monte Carlo rollouts."
In this new version, MCTS is not even used to evaluate a position! Speaking as a Go player, the ability for the neural network to accurately evaluate a position without "reading" ahead is phenomenal (again, read the Nature paper last page for details).
Re: AlphaGo Zero: Learning from scratch
#266Earlier quoted context omitted.
The concept of locality is nothing but a human weakness in Go, the best AI must read the whole board with every move. EDIT: From the paper: "Surprisingly, shicho (“ladder” capture sequences that may span the whole board) – one of the first elements of Go knowledge learned by humans – were only understood by AlphaGo Zero much later in training" I'm surprised by the author's use of the word "Surprisingly" here.
AlphaGo is still based around layers of 3×3 local convolutions. That represents a strong assumption about locality in the network design. I would expect AlphaGo to perform poorly on the game "Go with the vertices randomly permuted".
Well they didn't use inception. If their inception units have, say, a 7x7 conv, then ladders will probably be found much earlier.
Re: AlphaGo Zero: Learning from scratch
#267this is quite an achievement that it plays accurately. Does this have implementations for Strategy? No two games are alike, yet we are able to learn from our experience. Human's players be somehow comparing local positions from various games and deciding (sometimes wrongly) that they can be played similarly. Can it recommend not just individual positions, but general modes of play for classes of positions?
Re: AlphaGo Zero: Learning from scratch
#268Earlier quoted context omitted.
Agreed. I know I'm winning by 4 points but I have no idea about my probability of winning. However if I'm winning I know that I should play low risk moves and refrain from starting complicated fights. That increases the probability of winning. IMHO the exact value is out of reach for human beings.
It would be interesting to play human go, assisted by a go computer that doesn't say anything about moves, but rather just spits out, for each player, their current likelihood of victory if all further moves by both players were "what it would do." That way, each player could know, at all times, (one major factor that goes into) their probability of winning. They'd still have to mentally adjust it for the likelihood…
There is computer go program (maybe Crazy Stone?) that analyzes a game record and annotates it with the winning percentage for every move.
Knowing that the opponent's winning probability changed from 52 to 57 was interesting only because it hints at a mistake. In case of such a large change the program suggests the move it would have played.
I saw an annotated game record and there were no variations: I remember a suggested move that made me wonder "why!?".
Another benefit of seeing the value of the winning probability is an assessment of who's ahead. However that's already possible with the score estimation that programs and go servers provide. Sometimes is crude, sometimes is good, but it's the score, not the winning probability that humans can estimate when playing. The best probability estimate I can make is: if the score is close and the game is still complicated, it's 50-50; if the score is close but the game is almost over, it's 95-5 for who's ahead. If the score is not close, the player with more points will probably win.
Re: AlphaGo Zero: Learning from scratch
#269Earlier quoted context omitted.
I will be interested to see what kind of algorithms they have used to allow AlphaGo to learn from its own moves. Are these pretty generics algos or are these very customized and specific ones that only apply to AlphaGo and the game of Go?
They have a new reinforcement learning algorithm that should be generically applicable to anything where a long sequence of moves results in a specifically gradable outcome. > The neural network in AlphaGo Zero is trained from games of selfplay by a novel reinforcement learning algorithm. In each position s, an MCTS search is executed, guided by the neural network fθ. The MCTS search outputs probabilities π of playin…
Doesn't this sound very much like how a human learns to play the game? MCTS ~ play/experience (move probabilities); self-play with search ~ study/analysis (move evaluation); repetition and iteration to build intuition (NN parameters).