Breaking News
News & Analysis

Qualcomm Reveals Neural Network Progress

10/11/2013 03:40 PM EDT
24 comments
NO RATINGS
2 saves
More Related Links
View Comments: Oldest First | Newest First | Threaded View
<<   <   Page 2 / 3   >   >>
moloned
User Rank
Rookie
Re: chip technology?
moloned   10/14/2013 4:57:50 AM
NO RATINGS
"The brain is also very power efficient, he explained, consuming only about 20 watts at a cost of under a quarter of a cent per hour, whereas simulating the brain on a conventional von Neumann computer would take up to 50 times more power"

This statement is badly wrong and about 5-6 orders of magnitude off both in FLOPS and Watts!

This article reports an 83000 processor cupercomputer being able to deliver about 1% of the calculations performed by the brain and such super computers typically dissipate megawatts!

http://gizmodo.com/an-83-000-processor-supercomputer-only-matched-one-perc-1045026757

According to the German supercomputer centre in Juelich it will take an exaFLOP machine to simulate the entire brain in about 2020 with a power budget on the order of 20MW

Given current supercomputers can only manage about 10GFLOPS/W even this figure is in considerable doubt and would require almost 2 orders of magniture improvement in FLOPS/W in the next 10 years wit Moore's law and supply voltages plateauing

Kinnar
User Rank
CEO
Very Remarkable results obtained..
Kinnar   10/14/2013 5:20:22 AM
NO RATINGS
Very remarkable results are obtained by the Purdue University professor, it seems a really working technology, yet it is not much explained how the individual NPUs are working as neurons, but these technique will be having many roles to be played beyond it is explained in the article and explained by Qualcomm

moloned
User Rank
Rookie
Re: chip technology?
moloned   10/14/2013 6:26:46 AM
NO RATINGS
Of course a super-computer is not the only way to simulate the human brain and a domain-specific programmable SoC is one way of doing it.

At 34:40 at the end of Eugenio Culurciellos presentation on his vision SoC you can hear me asking a question about power dissipation of his system (I presented #movidius #myriad at the same conference)

The required 7500 chips in 2011 would dissipate a total of 15kW compared with the 25W power dissipation of the brain

Power is now down to about .6W according to his latest paper but this is still kW  

https://engineering.purdue.edu/elab/blog/wp-content/uploads/2011/11/pham-mwscas-12.pdf

This is a long, long way from what Qualcomm claim they can fit in a handset where 3-4W is the total for the complete smartphone including PA, baseband, WiFi/Bluetooth/GPS etc., display, Android OS and last but not least the budget for applications of 600-700mW

Robotics Developer
User Rank
Rookie
Re: chip technology?
Robotics Developer   10/14/2013 3:40:05 PM
NO RATINGS
I think that there are spiking neural network models availible on line somewhere (I remember seeing a link a while ago).  http://www.izhikevich.org/publications/spikes.htm

Has links to the models and some of the early work being done in NN.  Pretty cool if it really does work.  Looking forward to it.

MikeA..2
User Rank
Rookie
Wrong robotics law
MikeA..2   10/14/2013 3:49:18 PM
NO RATINGS
This may not be critically important in the overall scheme of things, but since others have called elements of this article into question it is worth noting that Asimov's Zeroth law is incorrectly described. The law that robots may not harm a human is captured in the first law. The Zeroth Law states that a robot may not harm humanity, which is quite a different thing. Anyone interested can get the lowdown at this Wikipedia page: http://en.wikipedia.org/wiki/Three_Laws_of_Robotics

 

mike

Jayna Sheats
User Rank
Manager
Neural network hardware
Jayna Sheats   10/14/2013 7:57:21 PM
NO RATINGS
The comments about various libraries, compilers, and other standard digital lore are well off the mark with respect to neural networks.  The basic structure of neural function has been known for a long time, even if the details may await future generatiions.  It is analog through and through, and no digital approach will work.  John Hopfleld and colleagues showed three decades ago how extraordinarily efficient a simple collection of analog processors could be at a test case - solving the traveling salesman problem.  Operational amplifiers were all the electronics needed.

So it can certainly be done in CMOS, but it will not be digital, and it will not involve a lot of C++...

DrQuine
User Rank
CEO
Neural nets and experience stores
DrQuine   10/14/2013 10:51:58 PM
NO RATINGS
The use of neural nets and "experience stores" allowing users to download expertise into their consumer products is a fascinating topic. I remember downloading neural net software for my MS-DOS computer in the early 1990's. It was the emerging breakthrough technology for solving image recognition problems. Obviously biological neural nets have served mankind very well (our brains). That said, there is an area of concern to me. The very fact that we don't know what cue is being used by the network to solve a problem may be the source of a serious issue. The software cannot be debugged or "certified" as correct. Imagine that we use a neural net algorithm to distinguish apples from oranges. Perhaps it does perfectly. Not knowing the "factor" it is utilizing, we might get into very unexpected results under unexpected conditions. Is it using color? If so, did we remember to test with green and yellow apples as well as red ones? Is it looking for stems? Will it mistake round candles for apples? When the variables being used are understood, a simple traditional program can run with less resources (and potentially less errors in a comprehensive test) than the neural net. I would predict that downloading a variety of "experience" updates to a neural net may result in some unexpected results.

Etmax
User Rank
Rookie
Re: NPU for automotive?
Etmax   10/15/2013 12:13:56 AM
NO RATINGS
I agree, neural has been on the boards for decades, Motorola had one about 15-20 years ago but no one ever seems to get it off the ground because they try to "program" them instead of letting them learn. Also power consumption has been a problem because they need to be essentially analogue and massively parallel to do real work (see our brain) and that doesn't translate well into silicon. I think some quantum deriviative of a neuron is going to be the real solution.

Etmax
User Rank
Rookie
Re: The role of federal research funding
Etmax   10/15/2013 12:16:32 AM
NO RATINGS
Sort of the Xerox or Bell of the 2000's :-) You'll see, that the government will declare them a monopoly (like Bell then) and the money won't be there anymore :-(

LarryM99
User Rank
CEO
Re: NPU for automotive?
LarryM99   10/15/2013 6:52:45 PM
NO RATINGS
One of the significant questions with neural networks is whether learning is continuous or not. This came up years ago when people started thinking about using them for control systems. If learning is disabled in an operational system then its utility is certainly limited, but if it is enabled then you run the risk of it learning something that would cause it to give a wrong answer (in some waus these things are very much like people!). This may be less of an issue for an intelligent UI than a direct control system, but one way or the other we need to understand how it will react. Some of the best Asimov stories revolved around ambiguity of interpretation of the Three Laws for a good reason.

<<   <   Page 2 / 3   >   >>
Flash Poll
Radio
LATEST ARCHIVED BROADCAST
EE Times editor Junko Yoshida grills two executives --Rick Walker, senior product marketing manager for IoT and home automation for CSR, and Jim Reich, CTO and co-founder at Palatehome.
Like Us on Facebook

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
EE Times on Twitter
EE Times Twitter Feed
Top Comments of the Week