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

OpenAI Universe

universe.openai.com

Post reply on HN
Print view View original

111–120 of 139 posts

Re: OpenAI Universe

#111

Post by vowelless »

noobermin wrote: post #92
pakl wrote: post #88

Earlier quoted context omitted.

I take the point to be that there aren't "deeper" fundamental principles at play in these models. Tremendous progress has comes from simply tweaking of the numbers of layers, or how the feed forward to each other (skipping layers, etc), or by throwing more computer power or data at the same basic algorithm. Where might we look for deeper principles? One idea is to consider what brains do and how they might be doing i…

As a physicist judging from the outside, I share some your feeling. Are there general laws governing "learning"? Theorems? Are there "deeper" things to learn as humans? The thing is people in the field don't need heavy intuition or math. In some ways that's good (if you just want a result to utilize) and in others, it's bad (if you are a curious person).

> Are there general laws governing "learning"?

This has been my question too...

Based on my understanding, it all really boils down to probability, statistics and a few important theories like Vapnik-Chervonenkis which provides the mathematical foundation for what "learning" is, can we even learn from the given data and how well can we learn (VC dimension, etc).

But I would love it if someone can point me to or explain / derive from core first principles the concept of "learning".

Re: OpenAI Universe

#112

Post by jessriedel »

TPCrow wrote: post #75

jesus man the constant switching of inflection due to jump cuts and shitty jokes was irritating. This is the stuff I always hate about your videos, and in general any "how to code an AI that does ___ with only < 10 lines of python". No you're giving me an extremely generic framework, and some vague direction that i can read off the linked page (the 9 lines), not how to actually write all these things, those are like…

You could have made these criticisms without the hugely inflammatory language.

Re: OpenAI Universe

#113

Post by pj_mukh »

flaviojuvenal wrote: post #52

Related but slightly off-topic, there is a great sci-fi story by Ted Chiang (the same author who made the story behind Arrival film) about humans raising AIs in an artificial world. The premise is that if we want AIs to act like humans, we must teach them like we teach humans: http://subterraneanpress.com/magazine/fall_2010/fiction_the_...

I'm getting a weekly dose of Ted Chiang by just farming links to free stories people post on HN! :D Thanks!

Re: OpenAI Universe

#114

Post by louprado »

TPCrow wrote: post #75

jesus man the constant switching of inflection due to jump cuts and shitty jokes was irritating. This is the stuff I always hate about your videos, and in general any "how to code an AI that does ___ with only < 10 lines of python". No you're giving me an extremely generic framework, and some vague direction that i can read off the linked page (the 9 lines), not how to actually write all these things, those are like…

Making a technical video for a general audience is real challenge. This video succeeds in that regard.

The audio cuts are likely because he had a deadline that didn't allow him to completely re-record the audio. I appreciate he inserted clips that added clarity despite knowing that he'd get negative comments for that effort.

Thanks for creating this video and sharing.

Re: OpenAI Universe

#115

Post by eli_gottlieb »

pakl wrote: post #97
noobermin wrote: post #92

Earlier quoted context omitted.

As a physicist judging from the outside, I share some your feeling. Are there general laws governing "learning"? Theorems? Are there "deeper" things to learn as humans? The thing is people in the field don't need heavy intuition or math. In some ways that's good (if you just want a result to utilize) and in others, it's bad (if you are a curious person).

In a sense, I would say yes there are learning laws, but it's still early in codifying them. Along one axis, you could compare: supervised, semi-supervised, self-supervised and unsupervised learning. Along another axis, consider that there are versions of each method that take into account temporal/dynamic data, versus others that require randomly shuffled static data. In the current problems of visual perception, I…

>In the current problems of visual perception, I think the field would benefit greatly a shift to focus on multiscale interaction/dynamics rather than on (static) statistics as it is currently (for more on this, see my colleague's blog: [1]).

Your friend's blog has a lot of good insights that I've seen in the theoretical neuroscience and computational cognitive science literature as well. Where do you guys work?

Re: OpenAI Universe

#116

Post by iotb »

Does OpenAI Universe communicate in any way with OpenAI remotely regarding activity in OpenAI Universe? Essentially, are there any call-home aspects to the code base? Or, is it possible to run this locally without any outside communication?

If there is remote communication, can you detail why and where it exists in code?

Re: OpenAI Universe

#117

Post by gdb »

iotb wrote: post #116

Does OpenAI Universe communicate in any way with OpenAI remotely regarding activity in OpenAI Universe? Essentially, are there any call-home aspects to the code base? Or, is it possible to run this locally without any outside communication? If there is remote communication, can you detail why and where it exists in code?

We don't call home. Once you've downloaded the Docker container, the only outbound network traffic should be downloading the requested SWF once for Flash games on demand (or for actually playing the game online, in the case of e.g. Slither). You can cache the SWF if you don't want it to be downloaded each time you start a new container.

Other than SWF downloading or specific Internet-enabled environments, running offline should just work.

Re: OpenAI Universe

#118

Post by gleglegle »

louprado wrote: post #114
TPCrow wrote: post #75

jesus man the constant switching of inflection due to jump cuts and shitty jokes was irritating. This is the stuff I always hate about your videos, and in general any "how to code an AI that does ___ with only < 10 lines of python". No you're giving me an extremely generic framework, and some vague direction that i can read off the linked page (the 9 lines), not how to actually write all these things, those are like…

Making a technical video for a general audience is real challenge. This video succeeds in that regard. The audio cuts are likely because he had a deadline that didn't allow him to completely re-record the audio. I appreciate he inserted clips that added clarity despite knowing that he'd get negative comments for that effort. Thanks for creating this video and sharing.

Agreed, I'd rather have the condensed information the video provided today than to wait for a more refined version.

Re: OpenAI Universe

#119

Post by cr4zy »

seasonedschemer wrote: post #37
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.

That would be the Berkeley DeepDrive project. http://deepdrive.io/ and http://bdd.berkeley.edu/ .

deepdrive.io creator here - I'm actually not affiliated with the Berkeley project of the same name. There's also a DeepDriving at Princeton plus plenty of other (mostly perception) projects using GTAV, so it can be confusing. I'm hoping the GTAV for self-driving car efforts can start to standardize around the Universe integration though. Having worked on it, I can say firsthand that the Universe architecture is definitely amenable to sending radar, lidar, controlling the camera, bounding boxes, segmentation, and other types of info that the various sub-fields of self-driving are interested in. Super-excited to see how people use it!

Re: OpenAI Universe

#120

Post by chronic92 »

state_less wrote: post #106
pakl wrote: post #89

Earlier quoted context omitted.

That's a great (and hard) problem! More generally, imagine AI that could learn the physics of the world. For example, if the ball is rolling away, the AI should be able to predict that the ball will look smaller on the next frame. Going further, if the ball is about to roll under a shadow, the AI should predict that the ball will become a darker shade of green. (After several years working in a robotics research comp…

Agree, it's not easy. Learning the basics, for example projecting a rectangle with 3d coordinates to 2d coordinates, then feeding the 2d coordinates into a NN and ask for the (depth) third dimension. Can you teach the NN a perspective transform? Can you rotate the rectangle and recognize rotation. Can you add other rectangles to the scene and detect each? Can you add color and lighting to infer more properties and ge…

This is a trivially easy problem if you have stereo cameras, just like humans.
Post reply on HN