This is a simple AI that learns how to spit an output from supervised data.
The relationship of each individual sets of patterns is not accounted for.
| Data Samples | Input | Expected Output |
|---|---|---|
| #1 | {1,0,1,1,0} | 1 |
| #2 | {0,0,1,0,0} | 0 |
| #3 | {1,1,1,0,1} | 1 |
| #4 | {1,1,0,0,1} | 1 |
| #5 | {1,0,0,1,1} | 1 |
{0,1,0,1,1} --------> OUTPUT = ?
Notice that the pattern of the EXPECTED OUTPUT is directly connected to the first integer of the INPUT pattern.
NOTE: The Pattern can change later on.
Therefore, OUTPUT = ? = 0.
// C is the nudge direction, and lambda is the nudge vaue
[
[50 sets of input combinations],
[193 sets of complex filter combinations]
]
- POSITIVE ----> Matters more
- NEGATIVE ----> Matters less
Weight Map Array
[
[ 193 Sets of relational synapses],
[ 2 sets of 193 synapses]
]
// Relational synapses are the connections between two neurons that are the same.
// Non-Relational synapses are undecided connections between two synapses.
[
[50 activations],
[193 activations],
[2 activations] // [0, 1]
]
[
[2 nudge directions],
// Nudge direction at index 0 => '0' connections
// Nnudge direction at index 1 => '1 conections
[193 nudge directions] // For each neuron in order
]
- Set up randomized pattern generation
- Be able to activate neurons and adjust weights
- Forward propagations
- Backwards propagations
- Iteration function
NOTE: I am looking for patterns so the filters/neurons are the patterns