[Ben.F.Rayfield]

Opensource GPL

Download from https://sourceforge.net/projects/physicsmata 2.0.0 and double click the file to instantly be playing with it. If anything goes wrong, try updating java.

    Cuttingedge research on automata and boltzmann AI. Has new visual kind of function editor where each click changes if a column is sine/sigmoid/half/double/exp/etc or which of the 0, 1, or 2 columns to the left are its params. Every click changes the patterns of waves seen on the left, which run that function you visually built at each point. The cellular automata API is, at each 2d point can be a different function and sizes of circles measuring average brightness around it. As we know from SmoothLife variation of Conways Game Of Life, these kind of automata can be Turing Complete, can calculate anything thats possible to calculate depending on the functions you visually build and TODO how boltzmann AI is hooked in. A 2d space tree optimizes the averaging of brightness in any chosen circle to cost only its perimeter not its area, so its really fast. - Also demo of a boltzmann machine learning mnist ocr dataset as you watch weights change and paint onto it and it dreams back at you.

Click the thumbnailed picture on the left. It looks like nothing you've ever seen, just 1 of the many automata functions you can build in a few seconds. And this is only the beginning. Boltzmann AI and painting automata functions like new colors onto an interactive video could easily be made with this foundation.

I'd like to hear your thoughts on this new kind of function building tool, and I know its very basic and could use some labels and dragAndDrop to save and reuse what you build, but I just got this working enough to prove its something new to the world. Geoffrey Hinton doesnt show you the edge weights changing while boltzmann learns or let you draw onto the AI's dreaming space as it dreams back at you. But of those basic parts that are something bizarrely new, what do you think?

[ranch vermin]

a boltzmann will only give you back the exact frames you put in, similar to a video tape, and accessable only with the data itself.

Thats actually a fact with every machine learning system,  do with it what you will, think it doesnt do anything because im not dreaming already....  your wrong - its possible to make a crazy intelligent computer just with that!

The cool thing you can do with one these 'frame accessable frame stores' is taking an errored frame, and bringing back the true frame.   (hopfields show this in a much easier to understand way, than listening to Hintons strange calculus vocabulary, or any of ours, especially if we invent to many of our own usages of words.) 

https://www.youtube.com/watch?v=fCvQcNzUZf0

now know what oscillation looks like, and it stored 2 frames restored from a different balance of noise.   so we need 100 million continuous frames, and pixel perfect recovery, then your a little closer to what you actually need.

My latest idea, is I can look at so many of these 'close photos' and produce a 'new photo' from looking at these, purely methodicly.   Team that up with data invarience/instancing.  means a part of a key, can have as many dynamic members instanced over that input area going into the boltzmann machine or whatever kind of key/photo memory it is. (excuse me for sounding like a crazy lunatic.)     

then I think you can get the working start of it.

Then you have to have a hard think,  how do I take my working vector robot, whos only ever occupying simple tasks, and get him to fire a bit whenever "someone laughs at a birthday party" or "im doing plumbing now" and then give him a->b symantic travel, tricky, isnt it,  then the evil future is here.