Hacker News Reader

Comment by Terr_ | Hacker News Reader

IIRC the flip-side was that it was hideously specific to a particular model and batch of hardware, because it relied on something that would otherwise be considered a manufacturing flaw.

View on HN

Not even one batch. It was specific to that exact one chip it was evolved on. Trying to move it to another chip of the same model would produce unreliable results.

There is actually a whole lot of variance between individual silicon chips, even two chips right next to each other on the wafer will preform slightly differently. They will all meet the spec on the datasheet, but datasheets always specify ranges, not exact values.

If I recall the original article, I believe it even went a step further. While running on the same chip it evolved on, if you unplugged the lamp that was in the closest outlet the chip the chip stopped working. It was really fascinating how environmentally specific it evolved.

That said, it seems like it would be very doable to first evolve a chip with the functionality you need in a single environment, then slowly vary parameters to evolve it to be more robust.

Or vice versa begin evolving the algorithm using a fitness function that is the average performance across 5 very different chips to ensure some robustness is built in from the beginning.

Yeah, if you took it outside the temperature envelope of the lab it failed. I guess thermal expansion?

There were also a bunch of cells that had inputs, but no outputs. When you disconnected them... the circuit stopped working. Shades of "magic" and "more magic".

I've never worked with it, but I've had a fascination with GA/GP ever since this paper/the Tierra paper. I do wonder why it's such an attractive technique - simulated annealing or hill climbing just don't have the same appeal. It's the biological metaphor, I think.

> slowly vary parameters to evolve it to be more robust

Injecting noise and other constraints (like forcing it place circuits in different parts of the device) are totally valid when it needs to evolve in-place.

For the most part, I think it would be better to run in a simulator where it can evolve against an abstract model, then it couldn't overfit to the specific device and environment. This doesn't work if the best simulator of the system is the system itself.

https://en.wikipedia.org/wiki/Robust_optimization

https://www2.isye.gatech.edu/~nemirovs/FullBookDec11.pdf

Robust Optimization https://www.youtube.com/watch?v=-tagu4Zy9Nk

Agreed, but to some degree relying on a simulator means it can no longer evolve truly novel solutions. No simulator would have accurately simulated many of the effects it was leveraging in the lab. Essentially you would only generate efficient uses of concepts we already know how to model / engineer well. Using the setup he used, it can generate things we don't understand well and can take advantage of. Or begin to study better.

That's was truly interesting about it to me.

long time ago, maybe in russian journal "Radio" ~198x, there was someone there describing that if one gets certain transistor from particular batch of particular factory/date, and connect it in whatever weird way, will make a full FM radio (or similar-complex-thing).. because they've wronged the yields. No idea how they had figured that out.

But mistakes aside, what would it be if the chips from the factory could learn / fine-tune how to work (better) , on the run..

AM radio can be "detected" with a semiconductor, so this kinda makes sense if you squint. If you can find it, someday, update this!

you can detect FM with an AM radio - "slope detection". This was in the back of my head when i typed this comment. A memory of picking up KFI on a tooth filling when licking the lid from a pudding when i was a kid. I had the real silver colored fillings. I used to get zapped, every time, and it was painful. It had to have been the semi-conductive amalgam, my arms acting as a dipole, and my skull and body the ground plane. But ancillary to that, i remembered the little electronic hobby kits we had, and a wire, a resistor, and a speaker was about all you needed to hear KNX 1070, a clear channel broadcast AM station in Orange County, CA.

At my highschool, we had FM radio transmitter on the other side of street. Pretty often you could hear one of the stations in computer speakers in library, so FM radio can be detected by simple analog circuits.

You need a PLL somewhere, because the audio signal is encoded in the frequency shift of the carrier (e.g. 98 MHz +/- 150 kHz).

Or maybe you were getting a leak from the audio signal before modulation, e.g. via power lines or something.

Interestingly, radios used to be called transistors colloquially.

Reply on news.ycombinator.com