Slashdot Mirror
Chips That Flow With Probabilities, Not Bits
holy_calamity writes "Boston company Lyric Semiconductor has taken the wraps off a microchip designed for statistical calculations that eschews digital logic. It's still made from silicon transistors. But they are arranged gates that compute with analogue signals representing probabilities, not binary bits. That makes it easier to implement calculations of probabilities, says the company, which has a chip for correcting errors in flash memory claimed to be 30 times smaller than a digital logic-based equivalent."
It would seem that they have reinvented the analog computer, but this time entirely on a chip. And probably (hopefully) with some logic that prevents errors due to natural processes like capacitive coupling.
12.5% that understands binary 87.5 that don't...
My -1 Troll is actually a +1 funny. And my -1 flame is actually a +1 insightfull.
So basically its a computer that makes up statistical computations and corrects them to fit the models on the fly? Lazy scientists, rejoice!
holy mother f*** s***! as a machine learning person, this is so exciting it got me tingly all over.
weinersmith
that this thing becomes a market hit ?
Been there, done that. Analog computers existed 50 years ago because digital computers were too slow. Even then, they were a nice market. Calibration is a big issue, and even with a perfectly calibrated machine you don't have a lot of accuracy. With the speed of today's digital computers, this is a (poor) solution in search of a problem.
Enjoy life! This is not a dress rehearsal.
The article mentions Bayesian calculations. Can these computers really speed up those calculations? Nowadays Bayesian calculations usually involve thousands of iterations of a technique called Markov Chain Monte Carlo (MCMC) unless the distributions in question are conjugate priors. The simulation then converges to the right answer.
The issue I see is that all these techniques are just math. They are either analytic (conjugate priors) or require strict error bounds in order get sensible answers (MCMC). There's no separate system of math that Bayesians use. Like many others, Bayesians just need quick reliable floating point mathematics. So anyway, I don't see how this can help Bayesian statisticians, unless it also revolutionizes engineering, physics, etc.
This is potentially a great advance. Everyone knows that analogue computing can greatly outperform digital computing (now each bit has a continuum of states so stores infinitely more data, each operation on 2 'bits'....you get the idea)....but there are many issues to resolve i.e.
1) Error correction - every 'bit' is in an erroneous state
2) Writing code for the thing - anyone got analogue algorithm design on their CVs?
This is the first step in creating an infinite improbability drive, you know...
I bet it's set to 'probably' blue screen, lol.
I can see how an AND gate would work.
Anyone want to guess how the others function?
Or am I on completely the wrong track here.
[Intentionally left blank]
Probability computing is not analog computing. Nor is it digital. Nor is it limited to error correction and search engines. It's a new implementation of a mathematical concept that allows arbitrary logic to be implemented smaller and faster than traditional digital chips.
Calling it analog is an insult that just shows you need to read a new book. ;-)
Yay Moore's Law! :-)
How much longer before we get the "infinite improbability machine"?
"Glory is fleeting, but obscurity is forever." - Napoleon Bonaparte
One step closer to the Infinite Improbability Drive (http://en.wikipedia.org/wiki/Technology_in_The_Hitchhiker's_Guide_to_the_Galaxy#Infinite_Improbability_Drive)
It's still just math. How will this be any different from digital calculations except for maybe the level of precision?
It's got absolutely nothing to do with analog computers. At all. The first application cited is even digital storage.
This would be ideal for mobile telephones and GPS devices. Signal reception, noise cancellation and error correction can all be done faster and with less energy when done in analog.
By Ben Vigoda, Co-Founder and CEO: http://phm.cba.mit.edu/theses/03.07.vigoda.pdf
Join the anonymous, help develop the network: http://www.i2p2.de
Analog computers sound so much more natural than digital ones..
It's vaguely familiar, but since no two circuits are *truly* identical at the analog layer, *and* change as the temperature changes, people used digital instead where 'mostly 0' is still '0' and 'mostly 1' is still '1' regardless. Otherwise you can't mass produce them.
Of more interest is people using analog-alike bitstreams, where the average number of 1's vs 0's in a random stream is the amplitude of the analog wave. They then blend the input streams together to produce the output stream. I've mostly seen this done by Royal Holloway University to produce neural chips that *don't* need squillions of interconnections - they just blend probability streams. Looks like people are playing with optical ones now too. Why not put a story up about that instead?
As long as a sperm whale and a bowl of petunia's don't suddenly flicker into existance when I power it on, I'm all for it.
First probability on a chip, next an improbability drive!
* Carthago Delenda Est *
...and out of those, 12.5% understand binary at any given time, and 87.5% do not...
Am I really the only person left that hates this construction? I know that it has become (very) common usage, but we, as nerds, should understand that details matter.
If one says that something is 50% smaller, we understand that to mean half the size. And if one says that something is 3000% smaller, or 30 times smaller, should we not understand that as not only taking no space, but actually giving us 29 times the original space back?
Unless we are making a three part comparison, which has new perils. If B is half the size of A, and C is 30 times smaller than B than B is to A, then we may understand the size of C as 0.5^30 times the size of A. However, if B is 99% of the size of A, then having C be 30 times smaller than B can mean that C is 70% of the size of A, or maybe C is 0.99^30 times the size of A.
Perhaps we should stick to saying what we mean, with things like "a chip for correcting errors in flash memory claimed to be one thirtieth the size of a digital logic-based equivalent"
See that "Preview" button?
Wow - just expand this idea infinitely, and we could have an infinite improbability drive. Think of the possibilities!
If it uses analogue signals internally then its an analog computer whatever those signals may represent at a higher level in the same way that a DSP is just as digital as a crypto chip even though the binary data is used for different things.
Ergo its an analog system. What those signals represent is irrelevant.
Those BSOD jokes were old 10 years ago. Did your time machine take a wrong turn and you ended up in 2010 instead of 1995?
why? can you elaborate?
If Pandora's box is destined to be opened, *I* want to be the one to open it.
You know what 30 times smaller means. In fact, you instinctively know it. Manufacturing the chip doesn't expand the universe by 29 times the size of a regular chip. We don't know ways to create more space so there isn't really any other way to interpret that expression. It conveys no misinformation... It is just silly to nitpick about that.
It is different than nitpicking about whether 10 is ten times larger than 1 or ten times as large as 1 because there is one correct one and another that conveys misinformation (though the difference isn't that relevant: those aren't usually meant to be exact statements to begin with). Here... It is just worthless.
Sounds like this hardware would be useful for Fuzzy Logic based AI applications. Fuzzy logic is useful for decision making and automating processes where multiple variables affect a range of possible reactions. Like when the cup you grab with your hand turns out to be very light because your girlfriend drank all your juice. When you initially grab the cup you start off with too much muscle activation and then adjust quickly at first then more slowly based on new sensory data. From common experience we know our grip strength isn't a function of one or zero but a range of activation that changes based on the ranges of other inputs. This is something Fuzzy Logic is good at and possibly something this chip would be good for too.
Just very definite ones, that is.
Ooops... make that Carver Mead, Addison Wesley 1989
Seastead this.
under my desk. After all, quantum mechanics is used inside it. So it's a quantum computer, right?
Robert "Prior Art" Heinlein strikes again.
http://en.wikipedia.org/wiki/The_Moon_Is_a_Harsh_Mistress
Take a look at the character "Mike."
Nice to see people making some more of his inventions into reality, even if this one is almost 50 years later.
Sounds like a neuron
... every couple of years people seem to "reinvent bull-sh*t" - which is what this is. I want reproducibility - same starting points give same results, not something that works some of the time... Who wants a to be on a plane that manages to land without crashing 90% of the time, and when it crashes, we just redefine it as landing properly but that everyone on board was DOA...
Yeah... I can see a market for that...
They sound warmer somehow...
And an analog component within a digital computer does not an analog computer make -- they all have analog components anyway, for fuck's sake.
Analog computers were cost-effective when a "floating-point option" came in a 5' cabinet and cost more than a luxury home.
ICs and cheap memory fixed that problem and analog calculation went the way of the dodo.
I'm a Programmer. That's one level above Software Engineer and one level below Engineer.
Google Maps