Artificial Neural Networks (ANN) are the supervised learning techniques whose logic is similar to biological neural systems. A simple ANN technique is the single-layer perceptron and it is a classification technique estimating a binary attribute whose value can be 0 or 1. The perceptron works like a neuron in the sense that it sums the impact of all the inputs and outputs to 1 if the sum is above a defined threshold. The model is based on the following parameters:
A weight for each feature, defining its impact
A threshold above which the estimated output is 1
Starting from the features, the model estimates the attribute through these steps
Compute the output through a linear regression: multiply each feature by its weight and sum all of them
Estimate the attribute to 1 if the output is above the threshold and to 0 otherwise
The models are as shown in the following figure:
In the beginning, the algorithm builds the perceptron with a defined set of coefficients and with a threshold. Then...