A Mind Made From Memristors 320
Csiko writes "Researchers at Boston University's department of cognitive and neural systems are working on an artificial brain implemented with memristors. 'A memristor is a two-terminal device whose resistance changes depending on the amount, direction, and duration of voltage that's applied to it. But here's the really interesting thing about a memristor: Whatever its past state, or resistance, it freezes that state until another voltage is applied to change it. Maintaining that state requires no power.' Also theoretically described, solid state versions of memristors have not been implemented until recently. Now researchers in Boston claim that memristors are the new key technology to implement highly integrated, powerful artificial brains on cheap and widely available hardware within five years."
Re:Quick question (Score:4, Informative)
The state is read by applying voltage to it, this voltage is less than the threshold required to change the state.
Neuromorphic CPUs (Score:5, Informative)
Even if the rest of the things explained in the article happen many years away, the last couple of paragraphs explain the trend:
Neuromorphic chips won't just power niche AI applications. The architectural lessons we learn here will revolutionize all future CPUs. The fact is, conventional computers will just not get significantly more powerful unless they move to a more parallel and locality-driven architecture. While neuromorphic chips will first supplement today's CPUs, soon their sheer power will overwhelm that of today's computer architectures.
The semiconductor industry's relentless push to focus on smaller and smaller transistors will soon mean transistors have higher failure rates. This year, the state of the art is 22-nanometer feature sizes. By 2018, that number will have shrunk to 12 nm, at which point atomic processes will interfere with transistor function; in other words, they will become increasingly unreliable. Companies like Intel, Hynix, and of course HP are putting a lot of resources into finding ways to rely on these unreliable future devices. Neuromorphic computation will allow that to happen on both memristors and transistors.
It won't be long until all multicore chips integrate a dense, low-power memory with their CMOS cores. It's just common sense.
Our prediction? Neuromorphic chips will eventually come in as many flavors as there are brain designs in nature: fruit fly, earthworm, rat, and human. All our chips will have brains.
Hopefully, this is the solution to 2018's problem of reaching atomic levels of miniaturization. We have a breaktrought to continue with Moore's law beyond current technology.
Re:Neuromorphic CPUs (Score:4, Informative)
I don't understand why most people focus on the maximizing transistor density part when 99% of applications call for minimizing cost.
Re:Neuromorphic CPUs (Score:4, Informative)
I think Moore's law is becoming increasingly pointless to most of the world. It talks about speed
It doesn't actually talk about speed at all; it talks about the cost of manufacturing chips of 2^n density where n increments every 18-24 months cost remains constant. It is, in fact, exactly what you go on to say is relevant despite the fact that what you're describing IS Moore's Law exactly.
Re:Artificial Brains? (Score:5, Informative)
That was the old theory. Years ago it was discovered that new brain cells are in fact produced throughout your lifetime.
Re:Solves the wrong problem (Score:4, Informative)
The process of figuring this out isn't going to occur magically. You need to test out your models at the systems level, with all the components working together. The more powerful the hardware we have to do this the more we can test and refine our models of how the brain achieves the same thing. This is both true if you're trying to model existing neuro architectures (like BU is) or if you're modeling evolutionary approaches like you describe above.
These memristive neuromorphic architectures hold the promise to get us orders of magnitude more processing speed while also keeping power levels low.
Re:Artificial Brains? (Score:5, Informative)
The reason slow replacent would work is because the opposite also works. Consider that everyday, hundreds of neurons in your brain are dying right now. Some are replaced by new neurons while others are not. And yet the continuity of you still exists.
Re:Artificial Brains? (Score:3, Informative)
And yet if we poke certain areas of your brain, we can not only make you do things, but you will also believe that you wanted to do them. Explain that.