Representing an Artificial Neural Network

Network.create [(1, 0.5), (2, -0.1)] [(1, 2, -0.5)]

create function

Network.create [(1, 0.5), (2, -0.1)] [(1, 2, -0.5)]

setValues is used to set values of input nodes

someNetwork |> Network.setValues [(0, 1), (1, -0.5)]

activate updates the network node values using the Step function

someNetwork |> Network.activate 

convert Network to String representation

someNetwork |> Network.toString

toDot create graph description see https://en.wikipedia.org/wiki/DOT_(graph_description_language) can be vizualised with https://rise4fun.com/agl or https://dreampuf.github.io/GraphvizOnline/ or using ports with https://github.com/mdaines/viz.js/

Network.create [(0, 1), (1, 0)] [(0, 1, -0.5)]
    |> Network.toDot
    -- == 
    digraph {
        0 [label="0=1"]
        1 [label="1=0"]
        0 -- 1 [label="-0.5"]
    }

fitness function defined as 1 - RMSE = 1 - Root of Mean Squared Error, this calcuation is done per sample and then summed for all samples, which should mean the result gives how many correct answer the network gave (kind of). see https://www.researchgate.net/figure/The-root-mean-squared-error-RMSE-when-the-genetic-algorithm-is-implemented-with-fitness_fig1_319382166

Network.create [(1, 0), (2, 0)] [(1, 2, 1.0)]
                    |> Network.fitness 
                        [ ([(1, 0)], [(2, 0)])
                        , ([(1, 1)], [(2, 1)])
                        ]
-- == 2.0  (Meaning both samples correctly answered)

get specific nodes with their current values