SLPs are are neural networks that consist of only one neuron, the perceptron. It can take in an unlimited number of inputs and separate them linearly.

Here is a small bit of code from an assignment I'm working on that demonstrates how a single layer perceptron can be written to determine whether a set of RGB values are RED or BLUE.

Of course the G could just be ignored, but this code is just to show how a SLP can be used to get rid of noisy data and find the correct answer

Here's some example training data

data = [#((R,G,B), CLASSIFICATION)
        [[0,0,255], BLUE],
        [[0, 0, 255], BLUE],
        [[0, 0, 192], BLUE],
        [[243, 80, 59], RED],
        [[255, 0, 77], RED],
        [[77, 93, 190], BLUE],
        [[255, 98, 89], RED],
        [[208, 0, 49], RED],
        [[67, 15, 210], BLUE],
        [[82, 117, 174], BLUE],
        [[168, 42, 89], RED],
        [[248, 80, 68], RED],
        [[128, 80, 255], BLUE],
        [[228, 105, 116], RED]
    ]

In this project my normalise() function takes in each of the input values and turns them into a value between 0 and 1. To do that I multiply each of the values by 0.003921568 because 1 / 255 ~ 0.003921568.

To modify the function for your own use, change out 0.003921568 for 1/(max input value).

Theoretically, this can be done by passing the desired number of inputs into Perceptron() when you create it (I haven't tested this yet).