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

How to build a robot that “sees” with $100 and TensorFlow

oreilly.com

Post reply on HN
Print view View original

51–60 of 63 posts

Re: How to build a robot that “sees” with $100 and TensorFlow

#51

Post by extesy »

Swizec wrote: post #7
taneq wrote: post #6

Earlier quoted context omitted.

Like so many other things in the field of AI, general object recognition was the "holy grail" because it was assumed that it required AGI. Now we've figured out a way to do general object recognition without AGI.

Whenever we figure out how to do something, we stop calling it AI or AGI. If the trend continues, will we eventually have a general AI, but won't consider it anything special? Will it have been just a small incremental step by then?

Yes, this is called AI effect: https://en.wikipedia.org/wiki/AI_effect

Re: How to build a robot that “sees” with $100 and TensorFlow

#52

Post by ebalit »

bonoboTP wrote: post #28
xiphias wrote: post #24

Earlier quoted context omitted.

Deep learning is the opposite of incremental. For a long time it was not clear whether/how we can learn multi layer networks efficiently. ImageNet changed everything.

Machine learning people basically agree that there weren't any big breakthroughs in deep learning. The success and the hype is mostly a combination of more computing power and more data. The algorithms (convolutional neural network etc.) were invented back in the 1980s and even earlier. There have been some improvements but they are incremental indeed. More use of ReLU, dropout etc. But it's not a new paradigm at all…

They weren't any recent breakthroughts. But LSTM and ConvNet are breakthroughts. It just took a lot of times to prove it.

Re: How to build a robot that “sees” with $100 and TensorFlow

#53

Post by Houshalter »

argonaut wrote: post #8
taneq wrote: post #6

Earlier quoted context omitted.

Like so many other things in the field of AI, general object recognition was the "holy grail" because it was assumed that it required AGI. Now we've figured out a way to do general object recognition without AGI.

Is there a story somewhere of AI researchers concluding general object recognition was the holy grail of AI? I get that a lot of people downplay achievements in machine learning by saying it's nothing like AGI, but it's almost a meme now that "once upon a time everyone thought that was the holy grail and they're moving the signposts" even when 1) nobody thought that, or 2) some people thought that and some people did…

Well I can't find any specific references, but I definitely recall getting that impression from old machine vision work. Decades of work to get models that were incredibly complex and hand crafted and barely worked. I don't know if I thought it would require AGI, but it would definitely require significant progress towards general AI.

Re: How to build a robot that “sees” with $100 and TensorFlow

#54

Post by bonoboTP »

ebalit wrote: post #52
bonoboTP wrote: post #28

Earlier quoted context omitted.

Machine learning people basically agree that there weren't any big breakthroughs in deep learning. The success and the hype is mostly a combination of more computing power and more data. The algorithms (convolutional neural network etc.) were invented back in the 1980s and even earlier. There have been some improvements but they are incremental indeed. More use of ReLU, dropout etc. But it's not a new paradigm at all…

They weren't any recent breakthroughts. But LSTM and ConvNet are breakthroughts. It just took a lot of times to prove it.

Convnets follow pretty naturally from multilayer perceptrons. Perhaps backpropagation was a breakthrough, enabling the training of ANNs on data, instead of hand-tuning.

But the idea of neural nets is very old, going back to Rosenblatt and connectionism.

Re: How to build a robot that “sees” with $100 and TensorFlow

#55

Post by ultrasounder »

dharma1 wrote: post #13
OilDerek wrote: post #12

Earlier quoted context omitted.

"I then built a simple Python webserver to spin the wheels of the robot based on keyboard commands that made for a nifty remote control car." So, not autonomous it would seem. With that and an arm, though, you could eventually get it to play fetch...

Ah, right. You could do this project with a cheap RC car and a phone running TF stuck on it

Since when did TF started to run on stock phones?

Re: How to build a robot that “sees” with $100 and TensorFlow

#56

Post by Florin_Andrei »

taneq wrote: post #6
rbanffy wrote: post #4

> recognizing arbitrary objects within a larger image has been the Holy Grail of artificial intelligence The Holy Grail is general AI. Recognizing objects is a side quest, perhaps a required step, but, by no means, the end goal.

Like so many other things in the field of AI, general object recognition was the "holy grail" because it was assumed that it required AGI. Now we've figured out a way to do general object recognition without AGI.

> Now we've figured out a way to do general object recognition without AGI.

I'm pretty sure we'll eventually learn to do anything without AGI, as a narrow task.

The trick with AGI is putting all those little things together. Perhaps that's the actual recipe for it, somehow. Turtles all the way down, who knows how many levels.

Re: How to build a robot that “sees” with $100 and TensorFlow

#57

Post by dharma1 »

sidarape wrote: post #43
dharma1 wrote: post #34

Earlier quoted context omitted.

yep, that's good if you want to retrain a pre-trained model for specific categories on new image data (that are pretty close to imagenet type images)

What do you mean by "pretty close"? What would be "not close"?

Something that looks very different to the ImageNet dataset. Microscope, ultrasound, satellite, x-ray etc images

Re: How to build a robot that “sees” with $100 and TensorFlow

#58

Post by dharma1 »

ultrasounder wrote: post #55
dharma1 wrote: post #13

Earlier quoted context omitted.

Ah, right. You could do this project with a cheap RC car and a phone running TF stuck on it

Since when did TF started to run on stock phones?

A few months ago.

https://github.com/tensorflow/tensorflow/tree/master/tensorf...

https://github.com/tensorflow/tensorflow/tree/master/tensorf...

Re: How to build a robot that “sees” with $100 and TensorFlow

#59

Post by MrQuincle »

argonaut wrote: post #8
taneq wrote: post #6

Earlier quoted context omitted.

Like so many other things in the field of AI, general object recognition was the "holy grail" because it was assumed that it required AGI. Now we've figured out a way to do general object recognition without AGI.

Is there a story somewhere of AI researchers concluding general object recognition was the holy grail of AI? I get that a lot of people downplay achievements in machine learning by saying it's nothing like AGI, but it's almost a meme now that "once upon a time everyone thought that was the holy grail and they're moving the signposts" even when 1) nobody thought that, or 2) some people thought that and some people did…

I - for what it is worth - would still say general object recognition, with the emphasis on general, is indeed the holy grail.

The ability to recognize objects like people do is not properly represented by current benchmarks. I can imagine that you can built a perfect robotic "bird spotter" but if you put that in a self-driving car I would not be surprised if it stops for something that's just a shadow, or if you put it on a humanoid it's unable to distinguish its own hand from that of its clone. Imagine two of them cleaning out the dishwasher. :-)

A lot of AI is still working only in lab conditions or restricted application domains. That's why I consider robots and cars so important in driving AI towards the "general" dimension.

Re: How to build a robot that “sees” with $100 and TensorFlow

#60

Post by sidarape »

dharma1 wrote: post #57
sidarape wrote: post #43

Earlier quoted context omitted.

What do you mean by "pretty close"? What would be "not close"?

Something that looks very different to the ImageNet dataset. Microscope, ultrasound, satellite, x-ray etc images

Ok, I see. Thank you.
Post reply on HN