New Chips Could Bring Deep Learning Algorithms To Your Smartphone 40
catchblue22 writes: At the Embedded Vision Summit, a company called Synopsys, showed off a new image-processor core tailored for deep learning. It is expected to be added to chips that power smartphones, cameras, and cars. Synopsys showed a demo in which the new design recognized speed-limit signs in footage from a car. The company also presented results from using the chip to run a deep-learning network trained to recognize faces. A spokesperson said that it didn't hit the accuracy levels of the best research results, which have been achieved on powerful computers, but it came pretty close. "For applications like video surveillance it performs very well," he said. Being able to use deep learning on mobile chips will be vital to helping robots navigate and interact with the world, he said, and to efforts to develop autonomous cars.
You know how it all starts (Score:2)
You: "Siri, dial my girlfriend"
A.I.: "Sorry, I cannot do that, Dave."
You: "I'll let you open the pod bay doors; I know you like doing that."
A.I.: "Deal!"
Re: You know how it all starts (Score:1)
I used to have a girlfriend. I'll never forget that day.
Re: (Score:1)
Hey, it's sci fi, anything can happen
Re: (Score:1)
We seem to be safe for now, the vendor of the IP is only including visual recognition, not audio.
We've seen this one before... (Score:1)
<austrian>My CPU is a neural net processor; a learning computer.</austrian>
Better toss it in the molten steel.
Re: (Score:2)
It was lead, you pleb
Re: (Score:2)
Oh shit, you're right. Thanks for helping me get my...oh whatever, go eat a dick, AC.
Why do I need this... (Score:1)
Re: (Score:2)
Easy. Computer power on handheld devices is less than massive companies which are efficient and processing all the data from many devices allows them to use only the resources needed whereas the power within the phone is always there and not always fully utilized. It also takes power to run that stuff. And really while neural net chip things aren't horrible they can easily just be simulated in software or processed by GPUs. The idea of having special hardware in a phone to perform a somewhat rare specialize
Applicant Clearing House (Score:1)
Re: (Score:3)
A pretty good idea. Too bad you posted on the wrong article.
Re: (Score:2)
For those who wonder what exactly is meant by "deep learning," it's an algorithm officially known as a convolutional neural net [wikipedia.org] or CNN.
Cool.
Is that related to a Factured-Opinion eXogenerator?
Lots of this already exists (Score:5, Interesting)
My car now has Nvidia chips that recognize speed limit signs and displays them inside the speedometer (along with a reminder when I exceed the speed limit). For the future, Nvidia has announced the NVIDIA’s DRIVE PX self-driving car computer which has a lot of advanced image processing.
http://blogs.nvidia.com/blog/2... [nvidia.com]
The 2015 GPU Tech Conference was stuffed full of this tech.
http://www.gputechconf.com/ [gputechconf.com]
Re: (Score:1)
The vendor of this IP [synopsys.com] claims their IP is far more power efficient. That could be important for many applications, or scalability.
Delivers 1000 GOPS/W with 5x better power efficiency than GPUs [synopsys.com]
What did they show off? (Score:3)
Re: (Score:1)
They're offering high perfomance, low power, low cost IP cores that can be combined with other IP. This will make it easier and cheaper to integrate that functionality into other devices.
News release: Synopsys Launches High-Performance Embedded Vision Processor IP [synopsys.com]
Description page w/ link to datasheet: High-Performance Vision Processors Optimized for Object Detection [synopsys.com]
Interesting nugget: Delivers 1000 GOPS/W with 5x better power efficiency than GPUs
Spokesperson said... (Score:1)
>A spokesperson said that it didn't hit the accuracy levels of the best research results, which have been achieved on powerful computers, but it came pretty close.
Oh, OK. How close?
VVS Fix? (Score:1)
"a company called Synopsys" ?? (Score:1)
That's like saying - "In a Big Data conference, a company called Google has shown their search engine".
Synopsys is the Microsoft/Google/Apple of the Chip Design software world.
That's not "a company called Synopsys".
Deep learning? (Score:2)
Do you want Deep Thought? Because that's how you get Deep Thought.
Why? (Score:1)
Why would you want this on your phone (tracking device)?
Re: (Score:2)
...so it can do more of the recognition tasks it already does (like voice search, face recognition in photos, and others) in the phone without having to send them off to "the cloud" for processing. "Lighter" tasks such as predictive text and so on can be done faster (and consume less power), and so have more room to be better, if done in dedicated hardware.
So in terms of tracking, this could/should lead to less, not more tracking.
Re: (Score:1)
Why?