OpenAI Universe • Hacker News Board

Re: OpenAI Universe

#41

Post by robryk » Mon, Dec 05, 2016, 10:09 AM UTC

tianlins wrote: post #35

jakozaur wrote: post #17

Browser tasks seems to be a greenfield field with amazing potential. What if AI can do anything what can human do you with a browser over the phone? Also love "bring your own Docker container format".

Well, if all GUI interactions can be automated, what would be our next human interface to computers/AIs?

Voice in one direction and voice and graphics in the other?

Re: OpenAI Universe

#42

Post by akrymski » Mon, Dec 05, 2016, 10:23 AM UTC

poppingtonic wrote: post #27

This is amazing! I was thinking of this problem when I saw a friend making a stop-motion video. The steps are super repetitive and I asked him, "maybe an DeepMind Atari-style RL agent can learn how to do this?" But I didn't want to do what DeepMind did to emulate Atari games with an Adobe editing tool. This is an experiment that I can now run.

Ha! Fascinating - I had the exact same thought when I saw a friend of mine make stop motion videos!

Re: OpenAI Universe

#43

Post by halflings » Mon, Dec 05, 2016, 10:25 AM UTC

poppingtonic wrote: post #27

This is amazing! I was thinking of this problem when I saw a friend making a stop-motion video. The steps are super repetitive and I asked him, "maybe an DeepMind Atari-style RL agent can learn how to do this?" But I didn't want to do what DeepMind did to emulate Atari games with an Adobe editing tool. This is an experiment that I can now run.

Never saw how stop-motion videos are edited. What's so repetitive? I thought that once you've taken all your pictures, you just put them sequentially and removed any frames that seemed off. Maybe you also need to decide how much time to leave every frame?

Re: OpenAI Universe

#44

Post by swah » Mon, Dec 05, 2016, 10:28 AM UTC

Layman question: isn't adjusting "hyperparameters" similar to writing a algorithm for playing a game, using human intelligence?

Related to the blog post: https://openai.com/blog/universe/

Re: OpenAI Universe

#45

Post by Cybiote » Mon, Dec 05, 2016, 11:43 AM UTC

This is astounding!

If requests are being taken, it would be useful to be able to search through the listed environments. And a poker environ for the internet section would be a good balance of fun, widely appreciable and a straight forward but very non-trivial environment.

Re: OpenAI Universe

#46

Post by poppingtonic » Mon, Dec 05, 2016, 12:06 PM UTC

halflings wrote: post #43

poppingtonic wrote: post #27

This is amazing! I was thinking of this problem when I saw a friend making a stop-motion video. The steps are super repetitive and I asked him, "maybe an DeepMind Atari-style RL agent can learn how to do this?" But I didn't want to do what DeepMind did to emulate Atari games with an Adobe editing tool. This is an experiment that I can now run.

Never saw how stop-motion videos are edited. What's so repetitive? I thought that once you've taken all your pictures, you just put them sequentially and removed any frames that seemed off. Maybe you also need to decide how much time to leave every frame?

This is true, but some are more complex, which is also due to the choice of tool. For example, [1], made by Chris King, using Adobe Premiere. This is a time-lapse of the process that he used to make parts of [2]. Notice the pattern that emerges when sequential images start lining up.

[1] https://youtu.be/M7Hr83OI-rs

[2] https://youtu.be/1-rFV_d6RH8

Re: OpenAI Universe

#47

Post by TeMPOraL » Mon, Dec 05, 2016, 12:41 PM UTC

iampherocity wrote: post #19

cing wrote: post #16

End game; I'd really like an AI agent for "in real life" tabletop games (like boardgames).

I call those friends.

Unfortunately, those "friends" have a lot of annoying issues that come with meatspace-produced wetware, and don't take (kindly to) pull requests.

Re: OpenAI Universe

#48

Post by vinay427 » Mon, Dec 05, 2016, 1:06 PM UTC

posterboy wrote: post #36

cpmsmith wrote: post #18

Earlier quoted context omitted.

I can't recall where, but I read that Tesla or Google were actually using GTA to train their self-driving cars, because it is a spectacularly advanced simulation of driving through an urban environment, so they didn't have to build their own.

>spectacularly advanced drivint gran tourismo is advanced, for a videogame at least

GTA is Grand Theft Auto, not Gran Turismo.

Re: OpenAI Universe

#49

Post by NamTaf » Mon, Dec 05, 2016, 1:25 PM UTC

From my initial reading, the end user can't create environments? Is that a feature that I can expect will eventually come?

Re: OpenAI Universe

#50

Post by tlb » Mon, Dec 05, 2016, 1:27 PM UTC

swah wrote: post #44

Layman question: isn't adjusting "hyperparameters" similar to writing a algorithm for playing a game, using human intelligence? Related to the blog post: https://openai.com/blog/universe/

It depends how many hyperparameters there are. Many popular general-purpose ML algorithms have only a handful of numbers for hyperparameters, so they don't embody much human input. And they can sometimes be tuned automatically.

Also, an algorithms that can learn 100s of different games with the same hyperparameters is more highly regarded than one that needs different hyperparameters for each.