From javilk at mall-net.com Wed Jul 8 15:32:38 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Wed, 8 Jul 2009 22:32:38 +0000 (GMT) Subject: [Mad-Scientists] A New Electrical Device Message-ID: <20090708223238.30954.qmail@mall-net.com> http://www.newscientist.com/article/mg20327151.600-memristor-minds-the-future-of-artificial-intelligence.html Folded: http://www.newscientist.com/article/mg20327151.600-memristor- minds-the-future-of-artificial-intelligence.html Transistor, Capacitor, Resistor... now, a memristor? A resistive device with memory, possibly enabling a 30 fold reduction in memory size. And one more thing, it's analog. We've just entered the digital revolution, changing the methodologies of the way we measure, record, even perceive the world. Digitization has spawned innumerable new tools and philosophical views. It has enabled general purpose devices, computers, to use algorithms and mathematics to process and predict reality. But digital is a finite approximation with quantization errors in handling an arguably analog world. 14 bit color in expensive cameras, 32, maybe 64 bits used in astronomical calculations... Shall we return from the wonders of digital-algorithmic processing to... With digital computers, we can iteratively calculate the slope of things, integrals, differentials... positioning our "points" on a very large variety of slopes and curves via calculations, iterative calculations. With analog... do we have the programmable iterative devices? Or must we rely on the slopes of discrete transformative devices? Do we have the ability to define algorithms of analog algorithms? Subroutining an the fly? Or are we stuck in hard wired discrete devices? I remember analog computers. Huge things with plugboards for programming (wiring) and ten turn pots for adjusting things. Sure, you could create fantastic filters for cleaning up the pop and hiss of vinyl-analog recordings, maybe even compensate a little for the non-linear behavior of mass-damped microphones and the vinyl itself. But at what cost? Digital made that easy. Marrying digital with analog... digital analog switches, pure digital analog switches, do not exist. Yes, relays; but relays have resistance on the contacts, and those go further and further astray, uniquely astray for each contact, the more you use them. Transistors? Again, device characteristics, slopes and such imposed upon the purity of the analog signal you are routing. So once again, we have operation induced errors. And that says nothing of thermal noise inherent in analog operation. But interesting thought... can we better approximate reality with digital, or analog? And where do we put the implementation boundaries of our preferences for digital vs analog analogies of the real world? http://www.newscientist.com/article/mg20327151.600-memristor-minds-the-future-of-artificial-intelligence.html Comments, discussion??? --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved. From celestialcognition at gmail.com Wed Jul 8 16:43:24 2009 From: celestialcognition at gmail.com (Benjamin Caplan) Date: Wed, 08 Jul 2009 18:43:24 -0500 Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <20090708223238.30954.qmail@mall-net.com> References: <20090708223238.30954.qmail@mall-net.com> Message-ID: <4A552F1C.2010908@gmail.com> Analog has a finite resolution just like digital does. The only real difference is that you can't do error checking in analog. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: OpenPGP digital signature Url : http://www.mad-scientists.org/pipermail/mad-scientists/attachments/20090708/7ddae85c/attachment.bin From javilk at mall-net.com Wed Jul 8 17:32:30 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Thu, 9 Jul 2009 00:32:30 +0000 (GMT) Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <4A552F1C.2010908@gmail.com> from "Benjamin Caplan" at Jul 08, 2009 06:43:24 PM Message-ID: <20090709003230.98965.qmail@mall-net.com> > Analog has a finite resolution just like digital does. The only real > difference is that you can't do error checking in analog. One of many, I think. Can we create the same kind of versatile processors for analog, as we do for digital? I mean, when you look at a transistor radio schematic, you see each device serves but one purpose in the chain. You have an IF section, Audio section, etc. The parts in each are different and generally not interchangeable. In contrast, when you take the same analog signal, digitize it and feed it in to a computer for discriminative processing to extract a signal, you see the same component used over and over and over again in each of the many mathematical operations. And then, the required interval being up, the operating system takes that whole processor and uses it to analyze stock market trades, format pictures to put on the screen, etc. before returning to process the next iota of sound. The same components being used in each of these seemingly very different tasks. If you like, you may change the size of your hard drive, shuffle your memory around, etc. yet it still works. The transistors in your CPU's multiplier section are functionally identical to those in the divide section, register transfer sections, etc. Can we create a general purpose analog processing device like we have in our computer processors??? --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved. From celestialcognition at gmail.com Thu Jul 9 15:13:42 2009 From: celestialcognition at gmail.com (Benjamin Caplan) Date: Thu, 09 Jul 2009 17:13:42 -0500 Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <20090709003230.98965.qmail@mall-net.com> References: <20090709003230.98965.qmail@mall-net.com> Message-ID: <4A566B96.20302@gmail.com> javilk at mall-net.com wrote: > Can we create the same kind of versatile processors for analog, as > we do for digital? > > Can we create a general purpose analog processing device like we > have in our computer processors??? I don't see the purpose, unless it's just to be able to say that you did it, like building a difference engine out of legos. And... possibly not. Or rather, such a device may not be practicably useful. Part of the power of general-purpose digital computers is their ability to re-process, re-interpret information indefinitely, through recursive depths and depths, vast processings. Analog devices, without the benefits of error checking, necessarily degrade the signal with every operation. Can you imagine an application in which it would be desirable to perform thousands of operations per second on analog signals? Telephone switching, perhaps. But that's one operation on thousands of signals per second. It's not really what computers are About. When would you want to perform thousands of successive operations per second on the same analog signal? -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: OpenPGP digital signature Url : http://www.mad-scientists.org/pipermail/mad-scientists/attachments/20090709/7a8ba733/attachment.bin From javilk at mall-net.com Fri Jul 10 00:01:13 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Fri, 10 Jul 2009 07:01:13 +0000 (GMT) Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <4A566B96.20302@gmail.com> from "Benjamin Caplan" at Jul 09, 2009 05:13:42 PM Message-ID: <20090710070113.10123.qmail@mall-net.com> > > Can we create the same kind of versatile processors for analog, as= > > we do for digital? > > Can we create a general purpose analog processing device like we=20 > > have in our computer processors??? > > I don't see the purpose, unless it's just to be able to say that you did > it, like building a difference engine out of legos. Smaller, purpose built devices, without the excess parts we tend to have when designing general purpose digital computers. > And... possibly not. Or rather, such a device may not be practicably > useful. Part of the power of general-purpose digital computers is their > ability to re-process, re-interpret information indefinitely, through > recursive depths and depths, vast processings. Yes indeed! > Analog devices, without the benefits of error checking, necessarily > degrade the signal with every operation. Can you imagine an application > in which it would be desirable to perform thousands of operations per > second on analog signals? Well in the sense, a radio does perform X-hz operations per second decoding the carrier signal. But it's in what we computer folks call a flow architecture, "data" flowing from one end to another with little feedback, except perhaps in the auto-gain control and auto-tuning. What was interesting, is that a group of researchers used programmable gate arrays (FPGA's) to try to evolve some sort of decoder, and ended up with a section which they could not understand, nor eliminate! Their conclusion was that there must be some principles in semiconductor technology which the research has not yet uncovered; but which evolution favored. I wonder if we just did. --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved. From celestialcognition at gmail.com Fri Jul 10 01:26:21 2009 From: celestialcognition at gmail.com (Benjamin Caplan) Date: Fri, 10 Jul 2009 03:26:21 -0500 Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <20090710070113.10123.qmail@mall-net.com> References: <20090710070113.10123.qmail@mall-net.com> Message-ID: <4A56FB2D.8010307@gmail.com> javilk at mall-net.com wrote: > What was interesting, is that a group of researchers used > programmable gate arrays (FPGA's) to try to evolve some sort of decoder, > and ended up with a section which they could not understand, nor > eliminate! Their conclusion was that there must be some principles in > semiconductor technology which the research has not yet uncovered; but > which evolution favored. I wonder if we just did. Heh heh heh. "The machine evolved in a way we couldn't have engineered. We don't understand it, even now. We know it does what we want; we just don't know why." I've read enough to know the signs of the coming rise of the machines. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: OpenPGP digital signature Url : http://www.mad-scientists.org/pipermail/mad-scientists/attachments/20090710/2437f8fc/attachment.bin From javilk at mall-net.com Fri Jul 10 10:13:41 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Fri, 10 Jul 2009 17:13:41 +0000 (GMT) Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <4A56FB2D.8010307@gmail.com> from "Benjamin Caplan" at Jul 10, 2009 03:26:21 AM Message-ID: <20090710171341.86358.qmail@mall-net.com> > > What was interesting, is that a group of researchers used=20 > > programmable gate arrays (FPGA's) to try to evolve some sort of decoder= > > and ended up with a section which they could not understand, nor=20 > > eliminate! Their conclusion was that there must be some principles in = > > semiconductor technology which the research has not yet uncovered; but = > > which evolution favored. I wonder if we just did. > > Heh heh heh. > > "The machine evolved in a way we couldn't have engineered. We don't > understand it, even now. We know it does what we want; we just don't > know why." > > I've read enough to know the signs of the coming rise of the machines. Kazynski was right; he was just an idiot when it came to expressing himself and promoting his ideas. As they say, even a stopped (mechanical) clock is right twice a day. (Does that mean our digital clocks are more stupid?) Over two decades ago, I was fiddling around with self organizing systems in my lab. I did not get very far; though I did create a non-semiconductor, non-computational problem solving type system oriented toward a specific (and very trivial) problem to try to prove some of my conjectures. In the course of this work, it occurred to me that the drive for homeostasis in complex systems equates with what we humans may call Love; and that Love is the most fundamental operational priority of neural type systems. Look, for example, at what dogs do for us just because we feed them. Or birds do for their neighbors, when a cat is on the prowl. And of course, all of the larger species do for their offspring. Evolution has insured love exists. In contrast, logic systems, such as what currently passes for artificial intelligence, have no such inherent operational priorities. One problem with such neural systems is their tendency to burn themselves in to a pattern of behavior, and then slowly burn, or wear, that behavioral apparatus to destruction. We also see this in humans under stress, such as where a manager learns how to cope with a tough work situation, and if he does not resolve the issue in some months (6 is often cited), ends up essentially perpetuating it. This is where upper management has to call in a a consultant such as myself to solve the problem, and usually orders the manager to take ALL of his accrued vacation days. This is where my work stalled; the thing just kept burning itself in. Far more equipment would have been necessary to inject additional signals to keep the patterns fluid enough to progress beyond elementary problems. One other problem, is that the self-organizing system can not be programmed; as all natural neural networks, the system must learn through trial and error. In most animals, this seems to be accomplished by the simple limitations of morphology. A baby horse soon learns to stand, then walk, and run; what else can it do with it's limbs? And so in a matter of a few days, it becomes a member of the herd. We, on the other hand, are too versatile with our upper limbs, and soon get into all kinds of trouble, such as pulling on tablecloths, ripping magazines, etc. So we end up building a far more elaborate and complex herd called a culture; and with those hands, build and add to what we call civilization. Takes us a heck of a lot longer to appreciate the limitations of our limbs, maybe 18 - 21 years; and the mind behind them, like maybe 40 - 60 years. I do wonder what kind of society the sub-species of mini-mastodon with four gripping lobes on the end of their trunks had. But I don't think they had a million and a half years to evolve their society and the brains behind it the way we did. They went extinct in the last ice age. We should put some emphasis on trying to understand what whales sing about. I think the contrast of their handless eden-like culture to our manipulation oriented culture could teach us a lot about our own culture. (I think eden is over-rated, because one only learns who one is by pushing the limits of what we may do, and allowing each to chose what and where to push. Hands allow so much diversity!) --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved. From celestialcognition at gmail.com Fri Jul 10 15:29:07 2009 From: celestialcognition at gmail.com (Benjamin Caplan) Date: Fri, 10 Jul 2009 17:29:07 -0500 Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <20090710171341.86358.qmail@mall-net.com> References: <20090710171341.86358.qmail@mall-net.com> Message-ID: <4A57C0B3.9070205@gmail.com> javilk at mall-net.com wrote: > Over two decades ago, I was fiddling around with self organizing > systems in my lab. I did not get very far; though I did create a > non-semiconductor, non-computational problem solving type system > oriented toward a specific (and very trivial) problem to try to prove > some of my conjectures. > > This is where my work stalled; the thing just kept burning itself > in. Far more equipment would have been necessary to inject additional > signals to keep the patterns fluid enough to progress beyond elementary > problems. > > One other problem, is that the self-organizing system can not be > programmed; as all natural neural networks, the system must learn > through trial and error. In most animals, this seems to be accomplished > by the simple limitations of morphology. A baby horse soon learns to > stand, then walk, and run; what else can it do with it's limbs? And so > in a matter of a few days, it becomes a member of the herd. Offhand, it sounds to me like your system became brittle because you were giving it too simply consistent a problem-set -- 'oriented toward a specific problem'. I strongly believe that generalists can only evolve under conditions that change faster than evolution can adapt to those changes. As soon as the circuit starts mostly-kinda-working, inject more variety into the definition of what you want it to do. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 197 bytes Desc: OpenPGP digital signature Url : http://www.mad-scientists.org/pipermail/mad-scientists/attachments/20090710/9e6bdd67/attachment.bin From javilk at mall-net.com Fri Jul 10 18:28:25 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Sat, 11 Jul 2009 01:28:25 +0000 (GMT) Subject: [Mad-Scientists] A New Electrical Device In-Reply-To: <4A57C0B3.9070205@gmail.com> from "Benjamin Caplan" at Jul 10, 2009 05:29:07 PM Message-ID: <20090711012825.30659.qmail@mall-net.com> > > Over two decades ago, I was fiddling around with self organizing=20 > > systems in my lab. I did not get very far; though I did create a=20 > > non-semiconductor, non-computational problem solving type system=20 > > oriented toward a specific (and very trivial) problem to try to prove=20 > > some of my conjectures. > >=20 > > This is where my work stalled; the thing just kept burning itself = > > in. Far more equipment would have been necessary to inject additional = > > signals to keep the patterns fluid enough to progress beyond elementary= > > problems. > Offhand, it sounds to me like your system became brittle because you > were giving it too simply consistent a problem-set -- 'oriented toward a > specific problem'. I strongly believe that generalists can only evolve "Problem set" is almost too strong a term for what kind of elementary system I had... > under conditions that change faster than evolution can adapt to those > changes. As soon as the circuit starts mostly-kinda-working, inject more > variety into the definition of what you want it to do. It was more like an organizational vs random type thing, and the "neural pathways", if they may be called that, which it used coalesced into one solidly used pathway. Given the few input leads and the lack of signal generation equipment... there was not much more that I could do to keep the "solution path" moving around in the neural-like matrix. A lot of things were missing in that early model, starting with time and money... another critical item missing was what one might call a discrete "write pulse" to confirm and enforce a successful path vs a non-successful path. But it was a fascinating look at how systems can, given a few properties, self-organize. One of our write pulses appears to be adrenaline. That explains a lot of human stupidity. We tend to go after anything which kicks our adrenaline up, regardless of whether it is harmful, risky, etc. We love to get scared and then tolerate mild injuries. As this sequence of adrenaline rushes progresses, you end up with the likes of Evil Knevil and lesser daredevils, soldiers, wars, empires. And a lot of dead bodies of those who were not coordinated enough, etc. All, it seems, because of a simple adrenaline based write pulse. I think the human body is less prone to producing adrenaline than other animals; because it takes vitamin C, something we do not produce, to produce adrenaline. And that may be one key to our intelligence -- we don't hit that write pulse as often as other animals; we don't Panic as often as other animals, and don't have the gram per gram muscle strength we so often see in other C generating animals. Not untill we get plenty of adrenaline flowing, can we exhibit the same superhuman strength we find in animals. So by writing less into our memories at the max level, we got smart instead of strong. --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved. From javilk at mall-net.com Fri Jul 17 09:21:30 2009 From: javilk at mall-net.com (javilk at mall-net.com) Date: Fri, 17 Jul 2009 16:21:30 +0000 (GMT) Subject: [Mad-Scientists] Cold Fusion Message-ID: <20090717162130.20926.qmail@mall-net.com> Intersting interview with Fleishmann of the team that brought us the wonder of cold fusion. He's retired, explains why they were forced to make their proclamations by the university -- so the university could gain the patents, rather than another university with other researchers whop were heading down the same path. X-rays, gamma rays, helium, other evidence found since their work now well supports some kind of fusion being responsible for the anomalous temperature spikes most see in cold fusion experiments; but no one has managed to really ramp it up in a controlled way to become a significant energy source yet. And no one really understands what inner mechanisms are responsible for the anomalous way the hydrogen fuses in cold fusion. http://www.newscientist.com/article/mg20327171.100-interview-fusion-in-a-cold-climate.html Our late member Mike Best tried one variant of cold fusion in his lab with me on the phone on the other side of the continent (in event of his demise), reported seeing "a lot of heat" on his instruments before the equipment destroyed itself. (Much as Fleishmann and Pons original equipment destroyed itself.) Sad my notes on the experiment burned with my lab last year. I'm still looking for a place to set up another lab somewhere "off the beaten path". Suggestions? Any of you tried your own cold fusion experiments? --javilk at mall-net.com---------------------------------- Life is to be LIVED regardless of what is out there. Fear destroys life. Destroy your fear and live. ------------------------------------------------------- Not to be construed as psychological advice. Void where prohibited by law. Not available in all mental states. ------------------------------------------------------- Another Javilk (tm) brand post. Copyright (C) 2007, Javilk at mall-net.com Copyright retained. All rights reserved.