The Powder Toy

Search For Powder Intelligence by Windspren

Windspren
Windspren
30 / 4
2nd Jul 2016
29th Sep 2016
The reborn SFPI! Our goal is to develop and identify forms of pseudointelligence and life within TPT in the forms of PSTN bots, computing systems, genome emulators, brain architecture simulations, and anything else we (or TPT AIs we find :D) can think of
sfpi tpte3

Comments

  • Dropelin
    Dropelin     5th Jul 2016
    While this is off the current topic, and is off the topic of electronics entirely, I find the behaviors of SOAP to be interesting. With different ctypes and life values it seems to form into different 'organisms', some balls of soap, some balls of soap that have tentacles, and some are flying lines of soap that form packs. Just a thought.
  • Schmolendevice
    Schmolendevice     5th Jul 2016
    "Weights," or current membrane potentials are mediated by the temperature of the HWSC alone. These are controlled by the inputting ARAYs. Synapse strength and or "biases" can be set by tweaking the temperature of those ARAYs. "Charging" is done by "hot ARAY" synapses being sparked by their INST "axons" and heating the HWSC whereby "cooling down" is mediated by ambient heat and is accelerated by "inhibitor synapses." Analyze this save where these components are isolated. id:id:1844423
  • QuanTech
    QuanTech     5th Jul 2016
    how are the weights of the perceptrons controlled? My idea includes using a controlled piston to push and pull batteries(weights).
  • Schmolendevice
    Schmolendevice     5th Jul 2016
    One must tweak around with the temperatures of the TSNS hillock and the inhibitory and excitatory ARAY "synapses" to get the desired inhibition effects etc. This system effectively accounts for both spatial and temporal membrane potential summation and uses TSNS to generate pulses and can theoretically handle arbitrary synapse counts and soma sizes. Reading the piecewise "perceptron function" on Wikipedia, my neuron component in ways acts like a temperature based perceptron.
  • QuanTech
    QuanTech     5th Jul 2016
    I'm just working on electronic artificial neurons. For now, it will be the perceptron
  • Schmolendevice
    Schmolendevice     5th Jul 2016
    On the right side, each of those small "neuron looking" blobs connected together with wires represent "neurons" more so than digital perceptrons. They use an HWSC soma and TSNS "axon hillock." It effectively produces "action potentials" with a maximum frequency of 7.5 Hz. ARAY represents either inhibitory or excitatory synapses and contributes to the temperature/membrane potential of the HWSC soma.
  • QuanTech
    QuanTech     4th Jul 2016
    +Schmolendevice Where are the perceptrons in that save? Besides, i am a bit better with the electronic version while you prefer organic ones. We'll see which ones are more efficient and compact. Thanks for coming to the rescue!
  • Schmolendevice
    Schmolendevice     4th Jul 2016
    @Kevino36 I'll do with a pastel vermillion RGB: [184, 53, 53].
  • Schmolendevice
    Schmolendevice     4th Jul 2016
    @QuanTech Monsieur, (I am technically self taught in computer engineering and watchmaking and am about to enter university for the former), I've seen you perceptron design, but my preference is in these more "organic" and flexible temperature based ones from id:1844423 as demonstrated in an old SFPI save id:1844947. Indeed I'm researching a bunch on neural networks from the biological perspective.
  • QuanTech
    QuanTech     4th Jul 2016
    this save now has more than 1k views!