User Tools

Site Tools


The Infinite Brain

In this project, I am undertaking the task of creating an infinitely scaleable brain. The idea is to create a structure that resembles an actual neural network in a brain that has been flattened and unrolled.

Biological Structure Overview

Neurons are made up of (essentially) four components: the Soma, axons, dendrites, and synapses.

  • Soma: stores the current charge of the neuron but also leaks charge over time. Once it charges up to a certain point, the charge is released via the axon hillock and sent down the rest of the axon to the synapses
  • Axon(s): the output trajectory (path?) of the neuron. Think of it as an extension cord from the Soma to the synapse
  • Dendrite(s): the input to the Soma. These tree-like structures collect and disperse the charge throughout the rest of the dendrites until it is collected by the Soma or dissipates.
  • Synapse(s): responsible for transferring the charge from one neuron to the next, usually via synapse-dendrite connection or even synapse-soma

Model Proposal

I don't need to simulate down to the exact proportions, so I propose building four elements:

  • Element_Soma:
    • Single element,
    • Very low movement probability or maybe even stationary.
    • When activated, it collects charge from all dendrites in the event window.
    • If the charge it collects is above a certain threshold, it tells the adjacent Element_Axon to “fire”.
    • If the number of times this atom “fires” within a certain number of activations, it dies and releases any Element_Axon and Element_Dendtrite attached to it
  • Element_Axon:
    • Chained together element
    • Very low movement probability or if it does move, it only moves “with” an Element_Soma or attached Element_Axon.
    • Only responsibility is to push a “fire” event to the next Element_Axon or Element_Synapse next to it (possibly integrate myelin sheathes into this element).
    • If not attached to either an Element_Synapse or Element_Soma, it dies.
  • Element_Dendtrite:
    • Chained together element
    • Very low movement probability or if it does move, it only moves “with” an Element_Soma or attached Element_Dendtrite.
    • Collects and perpetuates charge from stimulating Element_Synapse by pushing it in some direction along the attached Element_Dendtrites.
    • If not attached to an Element_Dendrite or Element_Soma, it dies.
  • Element_Synapse:
    • Attached to an Element_Axon
    • Very little movement or if it does move, it only moves with it's attached Element_Axon.
    • This looks for the closest Element_Dendrite and stimulates it by injecting the current charge it received from an Element_Axon.
    • If it is not attached to an Element_Axon, it dies.
    • May spawn new Element_Axon/Element_Synapse pair if it is constantly stimulated to strengthen the “learned” behavior.

Design Challenges

Given the previous information and stemming from the ever-present A-Life research cycle, some of my fears include:

  • Creating “attached” or “linked” elements. How to do this? Is it possible to do this?
  • Growing the network?!?! How can it grow? Each soma has an exponential decay function that determines if it spawns a new soma? Use Dreg/Reg?
  • Moving the network? Exponential decay function again? So maybe the outside edges “move” more than the internal edges?
  • Sensory neurons? Create separate elements designed to just look for things in their event window and have them stimulate themselves? Have other elements that stimulate them? I.E.: rods, cones and flash of light elements?
  • Output neurons? What can I do to interpret the output? WHAT DOES IT ALL MEAN?!?!?
  • Where can I find a u-shaped curve? Can I borrow yours?

Goals & Suggestions

  • Developing a network that grows would be rewarding in and of itself
    • Use DRegs and Regs to construct a robust network of semi-ordered structure
    • Soma elements consume regs to create new dendrites/axons
    • dentrites and axon elements consume regs to create branches
  • Trivial computation (nand gate?)
  • Nix Element_synapse
people/taylor_berger/biologically_inspired_nn.txt · Last modified: 2014/09/08 16:56 by tberge01