The Powder Toy

Brain evolution by Karakanlud

Karakanlud
Karakanlud
98 / 6
16th Sep 2015
18th Sep 2015
*updated with 5 synapse neurons*My neural grid simulations so far. The right one is a new construct, with improved neurons.
addprot addvirs bagels brain neuron neural grid system biology

Comments

  • Schmolendevice
    Schmolendevice     18th Sep 2015
    Sigh, once again a glorified pseudorandom chaos automata. When will folks learn to implement proper action potential based neurons on TPT? *More like this save is an innovation in its own ways, but some time in August I came up with a system for simulating simple temperature based 'perceptrons' which are connected in more realistic ways similar to a cortical column. (ogsterduck has addressed this I believe)
  • Karakanlud
    Karakanlud     18th Sep 2015
    whoops, *input
  • KetchupMaster
    KetchupMaster     18th Sep 2015
    You are very smart. I feel like its people like you who might end up becoming the next Ben Carson. :)
  • Karakanlud
    Karakanlud     18th Sep 2015
    (By evolution, i only meant the improvement of the grid ;_;) Yeah, real neurodevelopment is hardly possible in tpt. Also, the highest synapse count per neuron already made is five. But by diversifying the circuit, maybe we could make a grid with a certain relaxed signal pattern and reactions to imput, with other functions replaced by electronics, e.g. neural memory replaced with FILT memory units. Like an epic proto-brained cyborg t-100 ****** AI.
  • noobproducts
    noobproducts     18th Sep 2015
    lol i downloaded tpt for our school then deleted it because everyone was meesing arount with dest and sand.
  • greencarrson
    greencarrson     18th Sep 2015
    Holy Crap. The Powder Toy May Be the home of Many Future Scientists That Will be learned about in school one day. I mean seriously. We have created: Computers,Brains,Nuclear Power Plants And Even USB's!
  • killersor
    killersor     17th Sep 2015
    i was the one who suggested the Nble gas thing, well atleast i think i was the first one
  • RiftandRend
    RiftandRend     17th Sep 2015
    For this to be even slightly accurate, you will need about 7000 synapses per neuron.
  • ogsterduck
    ogsterduck     17th Sep 2015
    It is probably impossible to implement neuroevolution in TPT. Maybe just maybe you could do backprop with like 3 neurons simulated by a very good subframe computer using a static learning rate. Other than that it's impossible.
  • ogsterduck
    ogsterduck     17th Sep 2015
    As someone who has programmed several actual neural networks including training methons such as backpropagation and neuroevolution, this is NOT evolving. These can't even be called neurons.