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

Facebook AI Research Team Open Source DeepMask and SharpMask

code.facebook.com

Post reply on HN
Print view View original

51–60 of 89 posts

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#51

Post by visarga »

BOBOTWINSTON wrote: post #32
magicalist wrote: post #11

Earlier quoted context omitted.

> Is there any way for those of us with average intelligence to contribute to tools like this? The people doing this aren't magical geniuses; they've just put the time and work into the subject and have been able to get themselves into a position they can do this all day surrounded by others they can collaborate with. As with most human endeavors, the trick is to just get started, and not get frustrated and give up w…

I completed that course awhile back. Definitely a really great course if you can keep up. It is quite difficult, and I was quite terrible at it. I don't think I have any affinity for the topic, but at least now I can read/discuss the topic without being completely blind. There are also a lot of smaller components in the course that I found useful even not working directly in AI/ML. Just some general data modelling an…

Some AI professors recommend first jumping into a ML framework such as scikit-learn or Keras, which are more approachable, playing with them to the point where a little practical intuition will develop, and only then follow up with theory. Works better in practice than loading up on courses and math at the beginning, and helps the student form practical interest in the field and be more emotionally invested. In other words, first, hack your way in, and only then open the books.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#52

Post by visarga »

kmike84 wrote: post #35
said wrote: post #6

Is there any way for those of us with average intelligence to contribute to tools like this? I know I'm a decent developer, but I feel entirely entirely inadequate to participate in this enormous, scary world of AI.

ML can be an useful skill for developer. Think of it as of linux / sysadmin skills, or as web design / UI skills - they are all quite different from software development, but many software devs have them. I think that applied ML slowly becomes a similar thing - an useful tool in a toolbelt. It takes time to become really good at it, but basics are relatively easy to aquire. Libraries and learning resources are much b…

> but you don't have to know all state of the arts techniques to be dangerous

Often times you can be dangerous with just scikit learn, xgboost and Keras if you can do data prepping, pipelines and stacking/ensembles. Besides wonderful frameworks such as TF, Theano and Torch there are also "zoos" of ready made models, word vectors and such that can be used out of the box.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#53

Post by jupiter90000 »

dxbydt wrote: post #41
said wrote: post #6

Is there any way for those of us with average intelligence to contribute to tools like this? I know I'm a decent developer, but I feel entirely entirely inadequate to participate in this enormous, scary world of AI.

fwiw, over 4 years ago, I quit a risk quant job at an investment bank and joined Twitter. When I saw all the sophisticated ML systems they had deployed, I asked the same question - How do I, a person of average intelligence with zero exposure to ML, contribute ? Is there anything I can do to get involved? The answer I got was rather unpleasant and won't please everybody. No you cannot contribute. At least not in a di…

Thanks for that post, I'm in that waterboy place you're talking about and have been trying to find a way to the next level. If nothing else, sharing your experience was confirmation that helped me realize I'm not suddenly going to start doing ML at work if I'm paid to be doing other things.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#54

Post by visarga »

sushirain wrote: post #46
achr2 wrote: post #24

How 'deep' are the networks used in something like DeepMask, and how does it compare with the number of layers of the human brain?

For example, ResNet from 2015 had 152 layers. A real neuron takes in the order of 10ms to integrate and fire to the next neuron. Many subconscious reactions take less than 1 sec, which leaves time to a chain of length less than 100. Note that those neurons are not strictly arranged in layers. The human visual cortex has 10^12 synapses [1]. One popular 2015 deep learning net (ResNet 152-layers) used 10^12 FLOPs to cla…

It would seem that the current release from FB does top-down and bottom-up integration:

> To capture general object shape, you have to have a high-level understanding of what you are looking at (DeepMask), but to accurately place the boundaries you need to look back at lower-level features all the way down to the pixels (SharpMask).

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#55

Post by visarga »

tkinom wrote: post #47
sushirain wrote: post #46

Earlier quoted context omitted.

For example, ResNet from 2015 had 152 layers. A real neuron takes in the order of 10ms to integrate and fire to the next neuron. Many subconscious reactions take less than 1 sec, which leaves time to a chain of length less than 100. Note that those neurons are not strictly arranged in layers. The human visual cortex has 10^12 synapses [1]. One popular 2015 deep learning net (ResNet 152-layers) used 10^12 FLOPs to cla…

"It will take several years." A Noob question: If it reaches the "breadth" of human brain, how close will that to the "Skynet becomes self-aware" moment. What would it tell us about human, our society after it study, analyze millions, billions hours of FB, youtube videos?

As long as we do simple data processing, it won't become the dreaded Skynet. When we stuck a Reinforcement Learning system on top, feed it with reward signals and let it loose on the net or in real life robots, then it could become intelligent and develop its own goals/interests. So the Skynet part comes after reinforcement learning.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#56

Post by danielmorozoff »

dxbydt wrote: post #41
said wrote: post #6

Is there any way for those of us with average intelligence to contribute to tools like this? I know I'm a decent developer, but I feel entirely entirely inadequate to participate in this enormous, scary world of AI.

fwiw, over 4 years ago, I quit a risk quant job at an investment bank and joined Twitter. When I saw all the sophisticated ML systems they had deployed, I asked the same question - How do I, a person of average intelligence with zero exposure to ML, contribute ? Is there anything I can do to get involved? The answer I got was rather unpleasant and won't please everybody. No you cannot contribute. At least not in a di…

I would argue some of the most difficult problems in ml is data prep/ preprocessing.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#58

Post by cageface »

dxbydt wrote: post #41
said wrote: post #6

Is there any way for those of us with average intelligence to contribute to tools like this? I know I'm a decent developer, but I feel entirely entirely inadequate to participate in this enormous, scary world of AI.

fwiw, over 4 years ago, I quit a risk quant job at an investment bank and joined Twitter. When I saw all the sophisticated ML systems they had deployed, I asked the same question - How do I, a person of average intelligence with zero exposure to ML, contribute ? Is there anything I can do to get involved? The answer I got was rather unpleasant and won't please everybody. No you cannot contribute. At least not in a di…

I came to exactly the same conclusion about six ago when I decided to try to get into the field. I spent several months studying and implementing all the common algorithms by hand and working on various public data sets. I enjoyed it and ML is a fascinating subject, but once I started taking a closer look at the job market I realized that even big companies only need a few ML experts and that those roles tend to go to people with graduate degrees in the subject. So making the jump from web or mobile dev, for example, is pretty tough.

Re: Facebook AI Research Team Open Source DeepMask and SharpMask

#59

Post by fudged71 »

tectonic wrote: post #8

Future Facebook AR goggles will float people's names above their heads, replace billboard advertisements with live Facebook ads, and make people you 'mute' become invisible. It'll be a weird world.

Being colorblind, I'd be excited to point at an object and as what color it is. There are phone apps that can annotate individual pixel data, but the key is segmentation, average, and contextual/lighting cues.
Post reply on HN