Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AlphaGo Zero: Learning from scratch

deepmind.com

Post reply on HN
Print view View original

161–170 of 324 posts

Re: AlphaGo Zero: Learning from scratch

#161

Post by zardo »

tedunangst wrote: post #136
bradleybuda wrote: post #89

Earlier quoted context omitted.

In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. “What are you doing?”, asked Minsky. “I am training a randomly wired neural net to play Tic-Tac-Toe” Sussman replied. “Why is the net wired randomly?”, asked Minsky. “I do not want it to have any preconceptions of how to play”, Sussman said. Minsky then shut his eyes. “Why do you close your eyes?”, Sussman asked his teacher…

So Sussman was right the first time?

A random net has some random preconception. That doesn't mean it's a bad idea to try random preconceptions.

Re: AlphaGo Zero: Learning from scratch

#162

Post by maxerickson »

alex_stoddard wrote: post #114
kndyry wrote: post #104

Earlier quoted context omitted.

I don't think it's an overstatement to say that, since playing Lee Sedol in 2016, AlphaGo has completely revolutionized professional and amateur go. It's certainly not unprecedented — the last major revolution happened in the early 20th century (often called the 'Shin Fuseki' era [0]) — but AlphaGo has demonstrably surpassed any previous high-water mark. > I wonder if this system produced more new styles of play. Abs…

I have only skimmed the paper but one thing I don't see any discussion of is whether komi (the handicap given to white for going second) is correct. They do say the rules used for all games, including self-play, set komi consistently to 7.5 . If the strongest AI was consistently winning predominantly with one color it would be an indication that komi isn't fair for the best play. Of the 20 games released for the stro…

They could just test it right?

Or am I misunderstanding the hardware requirements?

Re: AlphaGo Zero: Learning from scratch

#163

Post by mannigfaltig »

jtolmar wrote: post #54
amelius wrote: post #48

Earlier quoted context omitted.

But I suppose they still do the searching/pruning with a separate piece of code (not a neural network).

Yes, but tree search + neural net is still pretty generic. It only assumes that you can enumerate branches.

It also presumes that one can simulate the world at low cost. In AlphaGo Zero it takes 0.4 s for 1.600 node extensions, but in this case the cost of the world is negligible. Anyway, assuming you need that many node extensions to get decent quality updates, that puts a rather a tight limit on the cost of simulating the world.

Re: AlphaGo Zero: Learning from scratch

#164

Post by waqf »

thefalcon wrote: post #133
panic wrote: post #36

Is AlphaGo Zero the first Go program without special code to read ladders? I'm curious how a pure neural net can read them, given how non-local they are.

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".

Re: AlphaGo Zero: Learning from scratch

#165

Post by durdn »

ericb wrote: post #7

I'm reminded of Eliezer Yudkowski's article "There is no fire Alarm for Artificial General Intelligence." Is this smoke? https://intelligence.org/2017/10/13/fire-alarm/ Yes, this is not an AGI. But the hockey-stick takeoff from defeats some players, to defeats an undefeated world-champion, to defeats the version of itself that beat the world champion 100% of the time is nuts . If this happens in other domains, like f…

You expressed my exact thoughts and I was about to link to the same insightful article. I guess my comment could've been shortened as a silent upvote, but I commented anyway.

Re: AlphaGo Zero: Learning from scratch

#167

Post by tim333 »

thefalcon wrote: post #132
farresito wrote: post #93

Earlier quoted context omitted.

I think he was cryopreserved, so he surely will be surprised once they wake him up in the future, assuming cryonics really works.

I'd certainly be surprised if I ever woke up from being cryopreserved. Which isn't to say that I'd object to the process if I had the disposable income and an understanding/cooperative family support structure, which I do not.

I wonder if it would be more popular if the cost was reduced to something similar to a regular funeral. It seems it might be a more cheery send off even if the chances of it working are questionable.

Re: AlphaGo Zero: Learning from scratch

#168

Post by arikrak »

ericb wrote: post #7

I'm reminded of Eliezer Yudkowski's article "There is no fire Alarm for Artificial General Intelligence." Is this smoke? https://intelligence.org/2017/10/13/fire-alarm/ Yes, this is not an AGI. But the hockey-stick takeoff from defeats some players, to defeats an undefeated world-champion, to defeats the version of itself that beat the world champion 100% of the time is nuts . If this happens in other domains, like f…

> But the hockey-stick takeoff

The hockey stick is lying horizontally though instead of vertically. If it took 3 days to go from 0 to beating the top player in the world, I wouldn't have expected it to take 21 days to beat next version. I guess something happens at the top levels of Go that make training much harder.

On another note, I didn't look at the details closely but it seems AlphaGo Zero needed much less compute training time than Alpha Go Master. Could getting rid of any human inputs really make it that much more efficient? That implies it will be able to have an impact in many different areas, which is a bit scary...

(Updated - it took 3 days to beat the top player in the world.)

Re: AlphaGo Zero: Learning from scratch

#169

Post by zardo »

braymundo wrote: post #53

I wonder if it could be applied to SC2 (since they already started the research: https://deepmind.com/blog/deepmind-and-blizzard-open-starcra... )

As is, no. There are too many possible actions and too little time between decisions.

I wouldn't discount it entirely though, some sort of clustering of actions may be able to reduce continuous action spaces to a manageable branching factor.

Re: AlphaGo Zero: Learning from scratch

#170

Post by gizmo686 »

arikrak wrote: post #168
ericb wrote: post #7

I'm reminded of Eliezer Yudkowski's article "There is no fire Alarm for Artificial General Intelligence." Is this smoke? https://intelligence.org/2017/10/13/fire-alarm/ Yes, this is not an AGI. But the hockey-stick takeoff from defeats some players, to defeats an undefeated world-champion, to defeats the version of itself that beat the world champion 100% of the time is nuts . If this happens in other domains, like f…

> But the hockey-stick takeoff The hockey stick is lying horizontally though instead of vertically. If it took 3 days to go from 0 to beating the top player in the world, I wouldn't have expected it to take 21 days to beat next version. I guess something happens at the top levels of Go that make training much harder. On another note, I didn't look at the details closely but it seems AlphaGo Zero needed much less comp…

This type of curve is what I would expect out of machine learning. At first there is rapid improvement as it learns the easy lessons. The rate then slows down as further incremental improvements become less impact.

What is, perhaps, surprising is that human play happens to be relatively close to the asymptote. Although this could be explained by Alphago being the first system to beat humans. If its peek performance were orders of magnitude higher than humans, a weaker program would have already beaten us.

Post reply on HN