McCulloch-Pitts Neuron
Table of Contents
A data structure that looks like the visual.
- \(d_1^{t-1}\) input at time step \(t-1\)
- \(w_1\) weight of input \(d_1\)
- \(\theta\) threshold/activation function
- \(X^t\) output of the neuron at time Xt
1. Restrictions/data types
- \(d_1\) can only be {0,1}
- \(w_1\) can only be {-1,1}
- \(\theta\) can only be [-inf,inf]
- \(X\) can only be {0,1}
2. Computation: input -> output
- sums up $d1w1,d2w2…$
- if the sum is smaller than \(\theta\), \(X^t = 1\)
- if the any input with weight
-1gives1, \(X^t = 0\)
3. Usage
3.1. simple logic gates
3.1.1. AND ATTACH
Only after all input gives one, can \(X\) be 1; other times 0
3.1.2. OR ATTACH
Any inupt be 1, \(X\) would be 1
3.1.3. NOT ATTACH
\(d = 1 \to X = 0\)
\(d = 0 \to X = 1\)
3.2. extended with time
3.2.1. register cell
3.2.2. memory cell
4. learning
no well-known algorithm for it
Backlinks
biocomputation
This page starts at topics covered in University of Liverpool’s COMP305 biocomputations.