REGISTER | LOGIN
Breaking News
News & Analysis

Neuromorphic Chip Market to Rise

Market to rise from ashes to billions
9/24/2015 09:43 AM EDT
31 comments
NO RATINGS
Page 1 / 2 Next >
More Related Links
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 4   >   >>
R_Colin_Johnson
User Rank
Author
Re: BCI Application
R_Colin_Johnson   10/18/2015 9:09:30 PM
NO RATINGS
Yes, there are many theories on this, Les, my patents are 93 and 90 and having to confront that issue every day, as will we all.

Les_Slater
User Rank
Author
Re: BCI Application
Les_Slater   10/18/2015 4:07:52 PM
NO RATINGS
I've given up on the issue of mechanical transference of the ego, it's not the most interesting question.

Les_Slater
User Rank
Author
Re: BCI Application
Les_Slater   10/18/2015 3:56:42 PM
NO RATINGS
My main point in my original statement, you live forever, whether you wish to, or not, is that there is a natural process which is much deeper and powerful than most suspect.

R_Colin_Johnson
User Rank
Author
Re: BCI Application
R_Colin_Johnson   10/18/2015 1:32:55 PM
NO RATINGS
Les, you're not alone. Many billionares are supporting "ego transference" methodologies, and of course Kurzweil has the Singularity Univeristy workig on it realization. The two big questions are when and whether it will only be available to the super-rich. 

Les_Slater
User Rank
Author
Re: BCI Application
Les_Slater   10/18/2015 11:42:58 AM
NO RATINGS
R. Collin,

Very difficult... and complex... question to answer. I'll start and then follow up.

Sometime, around when I was 10 or 11 (1954), with influences from maybe when I was 8, began thinking about machines being capable of supporting life, specifically sentience. Never questioned whether it was theoretically possible, seemed only natural.

Over the years the question was never quite central but often found myself giving it considerable thought. Never, ever questioned the theoretical. The big question was ego, whether or not it could support MY life... outside of my living biological body.  Very early concluded that a machine could be fooled into thinking it was me... and be quite satisfied that it was a continuation of ME.  The big problem with that is that I (capital I, EGO), would know better. After many years came up with a solution that would involve a transference of the EGO, very complex, but nevertheless, reduced to a technological problem.

With all that behind me, still some rough edges on the transference of EGO, been developing an understanding of just how central 'outside' influences are in determining just who we really are, as 'individuals'. Doing serious investigation of who I am, and how I got here, in part, lead me to conclude that we live beyond our biological bodies.

Les

 

Les_Slater
User Rank
Author
Re: Is The Era of Neurmorphic Computer Finally Here?
Les_Slater   10/10/2015 12:32:43 PM
NO RATINGS
realjjj, also, in my signal processing experience, I have found 'adequate' to be the catchword. In one case a brute force FIR solution would have been prohibitively expensive in terms of gate count and power. I used a mathematical relaxation methodology to find a minimal, not necessarily optimal, but adequate solution to minimizing FIR stages required.

Les_Slater
User Rank
Author
Re: Is The Era of Neurmorphic Computer Finally Here?
Les_Slater   10/10/2015 12:17:45 PM
NO RATINGS
realjjj, I see not only very great value, but necessity. In the sense we are discussing here, the quest for fidelity, or accuracy, that is a rigid comparing the detail of functionality with human organs, would be great folly and a hindrance to progress.

realjjj
User Rank
CEO
Re: Is The Era of Neurmorphic Computer Finally Here?
realjjj   10/10/2015 11:45:21 AM
NO RATINGS
If I may ask, where do you see approximate computing in this context?

Les_Slater
User Rank
Author
Re: Is The Era of Neurmorphic Computer Finally Here?
Les_Slater   10/10/2015 11:15:24 AM
NO RATINGS
Jayna,

I'm a very optimistic person, in a very broad sense. That doesn't mean I can predict the future, at least not in any detail with any great certainty. I learned a very long time ago, somewhere around third or fourth grade, that my views about the reality of nature (or the nature of reality), were often grossly wrong.

And also, on this particular question, you will note that I ended my previous post with 'As far as the question of the neuromorphic, we have yet to scratch the surface of topologies and algorithms that we will find in a couple decades.'

Your parenthetic, 'the human brain being 50W' makes me think that we are somewhat discussing the partitioning of the problem. Nothing particular in this discussion so far indicates that anyone is talking about duplicating the functioning of the human brain in any sort of autonomous sense. More specifically, certain functionality is proposed at various distributed points in three-dimensional space, and of course there are temporal implications.

The whole question that we are addressing is quite complex. And I particularly agree with ScribCore's, 'I like the philosophy that copying a bird is not necessarily the best way for man to fly, and copying the brain may not be the best way for computers to think.' I do believe that neural modeling has its advantages, but ultimately is crude... and will eventually run into scaling problems.

My very long association with electronic technology, going back to tubes, leads me to conclude that any problem that can be described logically, and / or numerically, is best implemented in the digital domain, converting any analog to digital as soon as possible, and convert back to analog, if necessary, as late as possible.

As far as not copying the bird, there will be shortcuts in functionality found that will be much more realizable in the digital domain, some not practically realizable in analog.

Back to the bird again, one of the books that I read in high school, Claude Shannon's 'The Mathematical Theory of Communication, Bell Labs, 1948, has influenced me enormously. The take-home that most reverberates in my head is that much (most) of the information conveyed is NOT transmitted in the channel, it primarily depends on information stored, and commonly shared, by both the sender and receiver. The analogy here, albeit a stretch, is that much of our brain function is either hard-wired, or otherwise not changed in real time.

I have found in my experience in signal processing that often even if signals appear very complex, at first glance random, there is often an underlying regularity, a predictable pattern, underlying that appearance. Of course Nyquist comes to mind here, but such patterns can be of a higher order than just a bunch of sine waves. One of the techniques that I have successfully employed is looking at large amounts of representative samples and computing solutions offline then applying the solutions to digital implementation. I believe much of the neural processing that we associate with brain functioning can be addressed by such methods.

Les

R_Colin_Johnson
User Rank
Author
Re: Is The Era of Neurmorphic Computer Finally Here?
R_Colin_Johnson   10/8/2015 9:05:36 PM
NO RATINGS
Thanks realjjj, I'll keep my eye out for new/better touchscreen innovations. I used to trave with a iPad and Bluetooth keyboard, but swithed to the new under 2 pound MacBook because of touchscree issues (still use tablet for infotainment, but not work) its also lighter than iPad+bluetooth keyboard and has bigger screen.

Page 1 / 4   >   >>
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed