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

Player of Games

arxiv.org

Post reply on HN
Print view View original

231–240 of 242 posts

Re: Player of Games

#231

Post by littlestymaar »

ahefner wrote: post #182
karpierz wrote: post #149

Earlier quoted context omitted.

That's actually the second mention, the first is when they introduce the games in section 4: > Today, computer- playing programs remain consistently super-human, and one of the strongest and most widely-used programs is Stockfish. They also go back to referring to it as Stockfish for the rest of the paper. An analogous situation in my mind would be if AMD released a new CPU and benchmarked it against an Intel CPU, on…

I'd be interested to see that benchmark. A ~3 GHz Pentium 4 sounds like a good reference point for single threaded performance since it's a reasonably modern OoO microarchitecture and reflects the moment that clock scaling stopped.

With a smaller cache, a less efficient branch predictor and only SSE for SIMD, I'd be curious to see the benchmark too but I'd be surprised if it was close.

I don't know if the RAM bandwidth being much lower would have an impact on CPU benchmark though.

Re: Player of Games

#232

Post by TaupeRanger »

kadoban wrote: post #226
TaupeRanger wrote: post #225

Earlier quoted context omitted.

It might play creatively, but it doesn't create any useful knowledge by doing so, making it kind of amusing but not the kind of creativity anyone is really interested in.

It creates useful Go knowledge. What else could be asked of it?

No it doesn't. Human Go players got worse after playing it. No one learned anything useful. The "knowledge" (if you can call it that) is all contained in an abstract, uninterpretable form that is of no use to anyone.

Re: Player of Games

#233

Post by kadoban »

TaupeRanger wrote: post #232
kadoban wrote: post #226

Earlier quoted context omitted.

It creates useful Go knowledge. What else could be asked of it?

No it doesn't. Human Go players got worse after playing it. No one learned anything useful. The "knowledge" (if you can call it that) is all contained in an abstract, uninterpretable form that is of no use to anyone.

Who got worse, and how? Seems like a pretty unlikely claim to me on the face of it.

Re: Player of Games

#234

Post by thom »

randomswede wrote: post #221
7thaccount wrote: post #145

Earlier quoted context omitted.

This was my understanding as well, but I might have read into it. The culture minds are in freaking hyperspace to get around lightspeed limitations on computations. He for sure can't beat that, but he could beat someone on another planet at their own game that he literally just learned in the year it took to get there. A game that permeates every aspect of their civilization. I do assume his drone could beat him as w…

One of the reasons that Contact (and Special Circumstances) have (some) humans[ ] around is for intuitive leaps. I can't say I recall which Culture book this is mentioned in, but it is, in one of them. [ ] let's go with "human" as a general term for Culture biological citizens, it is probably a bit incorrect, but gets the point across.

The 'referers' are in Consider Phlebas, a small group of humans among trillions who are able to reliably predict the future better than Minds. This would be one argument in Gurgeh's favour, but Banks later admitted how flimsy the idea was.

Re: Player of Games

#236

Post by Buttons840 »

jart wrote: post #100
jm547ster wrote: post #79

Earlier quoted context omitted.

3 years after...

Here's what Lee Sedol said when he retired: > With the debut of AI in Go games, I've realized that I'm not at the top even if I become the number one through frantic efforts. https://en.yna.co.kr/view/AEN20191127004800315 He'd been playing Go professionally for 24 years. I never said he ragequit. He's too great a man to do something like that. Lee instead apologized for his losses, stating "I misjudged the capabiliti…

He's the last human to ever beat the strongest Go AI. I don't know if he's happy about it, but he'll have a special place in the history books because of that. And like Chess, the game of Go will continue to be played and loved.

Re: Player of Games

#237

Post by Buttons840 »

mudlus wrote: post #69

Yawn, show me a computer that game make fun games

Solving the game comes before solving for fun. If we create an AI that can win, then we can hamper the AI in fun ways, or give it an altered objective function that maximizes the players fun.

Re: Player of Games

#238

Post by cygaril »

gwern wrote: post #227
fxtentacle wrote: post #207

Earlier quoted context omitted.

You can mathematically prove for a lot of different algorithms (including PPO, DQN, IMPALA) that given enough experience with the game world, they will eventually converge to the optimal policy. It's just that the "enough experience" part might be so large that it's practically useless. If I remember correctly, the DeepMind x UCL RL Lecture Series proves the underlying Bellman equation in this video: https://www.yout…

I don't think you can prove that (forgive me if I don't sit through a 2h video). Those all are susceptible to the deadly triad, and AFAIK there are no convergence proofs of any kind for the big model-free DL algs, and it would've been big news if someone had proved that a real-world version of PPO/DQN/IMPALA does in fact converge in the limit. Sutton's book and earlier proofs only cover cases where you drop the nonli…

Standard RL algorithms will converge to optimal play versus a fixed opponent, but will not find an optimal policy via self play.

One intuitive way to see this is that a sequence of improving pure policies A < B < C < etc. will converge to optimal play in a perfect information game like chess, but not necessarily in an imperfect information game like rock/paper/scissors where Rock < Paper < Scissors < Rock, etc

Re: Player of Games

#240

Post by CRConrad »

mrslave wrote: post #224
bewaretheirs wrote: post #181

Earlier quoted context omitted.

I keep hearing recommendations for the Culture books so I tried reading it recently and it just didn't work for me -- I gave up on it halfway through, which is rare for me.

Me too with Consider Phlebas. Then I hit the Alastair Reynolds novels pretty hard and now I'm stuck for new material. Dune is en vogue so perhaps that's the right read next? I really enjoyed Vernor Vinge's A Deepness in the Sky but couldn't quite get into A Fire Upon the Deep but it still sits on my shelf taunting me.

> read next?

Stephen R. Donaldson: He's more known for The Chronicles of Thomas Covenant, but you might prefer The Gap Cycle; it's more typical SF.

Also, the Night's Dawn trilogy by Peter F. Hamilton. Perhaps Dan Simmons? And of course anything by Charlie Stross.

Post reply on HN