[goaty]

Yes antennas lock! Figure that out and I'll use it for my project. Wireless adjustments would be super helpful.  I NEEDS DEM ANTENNAS!!!!!

I vote that's a good idea too.   it means you don't need to store the mesh for the perceptron, if you wirelessly connect to the retina instead.
you use a complex batch of filters?- for what eye pixels it stands for,  for its image.       that's actually a huge cheat if its any good.  if you cant
do rf,  you can just use an ORDINARY MICROPHONE! if the pixels emit a frequency each!

[HS]

you can just use an ORDINARY MICROPHONE!

Nice! I'll try to help if anyone decides to figure out large scale wireless communication. I tried to do it before with lights but ran into some difficulties... Seems like all of our projects could benefit from this kind of tech.

[goaty]

It actually is a cool idea.  a "wireless perceptron"


<spoiler alert>>>
so instead of doing the impossible to wire all to all mesh,  you need a filter block on each cell that picks out the frequencies each one wants.if it was hard programmed,  youd need a resistor and capacitor block, but they could be really small.   but if you wanted it to learn, the filters would have "tune into" the retina. 

[HS]

Alright, I'll take a look at filters on the weekend, gotta catch up on my non AI studies  .

[AndyGoode]

I vote that's a good idea too.

Add my vote for wireless neural networks, too. For my own reasons I believe that is the technology of the future.

[HS]

How about the power? Wireless electromagnetic coils and/or photovoltaic, or power supply wires and just go wireless with the communication?

[goaty]

   love this idea, HS.

That "photo voltaic" idea is a very good one as well, and H.S. definitely came up with that one. Because logic doesn't need high volts to run, I think it would work.  But when its coming out of the net to the robots legs then youll need to amplify it and maybe go to a solid battery connection then.   But maybe to retransmit after you receive might take some amount of power, I didn't think of that.

So that initial idea from ages ago, to have floating in a soup, picking up the signals from each other, might actually be practical,   just have to make sure the signal is getting from all the blocks to all the blocks without getting obstructed.  Then the cells just need a filter per wireless synapse.

They could be a chip board each,  I think if you wanted to make it more rigid, they could just slot in like cards, maybe 2cmx2cmxmm if you use really small low voltage parts, with the solar panel on them, picking up the light...  and maybe 2 piezoelectric transducers for the transmitter and receiver??

Anyhow I like the idea,   if I ever implement this myself ill put your callsign somewhere printed on the cells somewhere.

[AndyGoode]

So that initial idea from ages ago, to have floating in a soup, picking up the signals from each other, might actually be practical,   just have to make sure the signal is getting from all the blocks to all the blocks without getting obstructed.

Biological neurons already float in a 'soup' of sorts. It's called 'glial cells', and glial cells are like the packing material between those spindly neurons, and they also supply nutrients and oxygen to the neurons.

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

It shouldn't be too difficult to find some sort of battery acid, fuel, or other energy-supplying chemical to circulate through artificial glial cells to power the artificial neurons.

[goaty]

I wonder if that's where HS got the idea from.

Heres a picture of a net,  see how many wires there are?  if you could replace the wires with a filter in the cell itself for each connection, plus maybe a couple of piezos, the cells would be bigger but it would be a lot easier to fit it together, wiring it would be heaps harder, getting your fingers into that mesh would be impossible.

[AndyGoode]

Another idea would be to use the principle of a self-winding watch... Just put a tiny current generator at each neuron that generates a tiny bit of current when two surfaces move near each other, like tiny magnets or tiny pendulums, and just physically jiggle the whole network every once in a while to ensure that all those tiny generators keep supplying energy.

[goaty]

yeh excellent idea.   

use a mechanical oscillator and power em all at the same time from the outside.
if it were just weights the whole thing could work without electricity too, using solid,liquid or gas logic instead.

[HS]

Nice ideas!
Before, I thought that if the cells are suspended freely then the cohesion substance needs to be resistant to change, like something between jello and hand sanitizer. Otherwise everything would flow and disarrange with thermal, nutrient, and external motion currents. But it would be literally mind-blowing if it could do that without consequences. Before, it wouldn't have worked because the use of light signals made it direction & distance sensitive, but now I don't see why not, if the signals can reach everywhere. As you said, this would also simplify the wireless charging since probability would be working with us, each energy converter would eventually be in an orientation/motion to absorb power. ...So on second thought lets do that. Actual neuron soup! Yeah Baby! Chaos FTW!

[HS]

I wonder if that's where HS got the idea from.

I got my first idea by thinking about these symbols, lol:

[goaty]

I wonder if that's where HS got the idea from.

I got my first idea by thinking about these symbols, lol:

ha funny.

[goaty]

I was thinking about it again,  and if you went to a megahert, you could have a 1000x1000 screen connected to as many cells as you want (even billions) because they are only different combinations of the screen,  and if you wanted to do deep learning with it, maybe u could isolate the volumes, and go through some kind of redunction back to a million again, and keep going.

So even if it was a megahert only, I think its still quite workable to something amazing still - but ud have to play around with it a bit.