The First Evolving Hardware?

Masq666 writes "A Norwegian team has made the first piece of hardware that uses evolution to change its design at runtime to solve the problem at hand in the most effective way. By turning on and off its 'genes' it can change the way it works, and it can go through 20,000 - 30,000 generations in just a few seconds. That same number of generations took humans 800,000 - 900,000 years." The University of Oslo press release linked from the article came out a few days ago; the researchers published a paper (PDF) that seems to be on this same technology at a conference last summer.

Re:GA in hardware

[wish bot]

And it's been done before - at least once - http://www.newscientist.com/article.ns?id=dn2732 [newscientist.com] - there's another one too, but I can't find it right now. Crazy stuff though.

Misleading

[suv4x4]

At first glance, this is supposed to impress us with the hardware:

By turning on and off its 'genes' it can change the way it works, and it can go through 20,000 - 30,000 generations in just a few seconds. That same number of generations took humans 800,000 - 900,000 years.

In fact the simplest DNA based organisms/structures (bacteria, virii) have the shortest "life span". The number of generations per sec. isn't anything to brag about.

All complex organisms have some sort of lifespan longer than a microsecond. For a good reason: people pass on knowledge and adapt *during* their life span (not genetically of course, but our brain allows us to adapt a lot without such).

Hype aside, interesting development, but I wish those publications wouldn't use misleading statements in pale attempts to impress us.

Re:GA in hardware

[wish bot]

Ahh ha - found it - http://www.informatics.sussex.ac.uk/users/adrianth /cacm99/node3.html [sussex.ac.uk]

My favourite bit:

Yet somehow, within 200ns of the end of the pulse, the circuit `knows' how long it was, despite being completely inactive during it. This is hard to believe, so we have reinforced this finding through many separate types of observation, and all agree that the circuit is inactive during the pulse.

Crazy stuff indeed.

Re:Hardly new

[Teancum]

I never said it was easy, but I have even seen it mathematically proven that any algorithm can be done in hardware, and I've duplicated most hardware into software myself, for those designs that I wanted to emulate.

This is not just a very small subset of designs. It is a matter of cost and if the engineer wants to put forth the effort to implement the whole thing in hardware. Trying to convert a 1st person shooter game like Doom into a pure TTL logic would make the game very responsive and give you screen resolution to kill for, but would it be worth the engineering effort to do that?

Race conditions and other "bugs" have other causes that may be due to ineptness on the part of the engineer, or because you havn't really thought the problem through sufficiently. Or there may be other things to look at as well. But don't tell me you can't implement in pure TTL logic something like an MPEG encoder.... which is a very complicated mathematical algorithm. I can give you part numbers for MPEG encoders if you really want them in your next design, as they are commercially available.

There is nothing that would stop you from implementing in hardware something like a neural network either... oh and those are indeed implemented in hardware. They are usually done in software mainly because of the cost involved, and you can use a general purpose computer to perform experiments on them. Other adaptive software algorithms have also been implemented on both hardware and software for some time as well. As I said, this is very old news here with this article.

Evolvable Hardware is *very* old news

[feijai]

Evolvable Hardware is so old it's got its own acronym (EH), it's own wikipedia entry [wikipedia.org], and its own conference [nasa.gov]. In the early '90s a researcher (I forget the name, oops) was using a GA to evolve circuits for an FPGA, which were tested on the FPGA and an oscilliscope directly to assess their fitness. NASA's done lots of evolvable hardware: in particular antenna designs which have flown in space. And there's a whole subfield of evolvable modular robotics.

And if we're talking about hardware simulation, the first significant use of evolutionary computation (GAs etc.) was Larry Fogel's work on evolving finite state automata machines in the 1960s. In the 1990s John Koza was using genetic programming to evolve patentable computer circuits in SPICE.