Learning OpenCV 4 Computer Vision with Python 3 - Third Edition

Understanding neurons and perceptrons

Often, to solve a classification problem, an ANN is designed as a multi-layer perceptron (MLP), in which each neuron acts as a kind of binary classifier called a perceptron. The perceptron is a concept that dates back to the 1950s. To put it simply, a perceptron is a function that takes a number of inputs and produces a single value. Each of the inputs has an associated weight that signifies its importance in an activation function. The activation function should have a nonlinear response; for example, a sigmoid function (sometimes called an S-curve) is a common choice. A threshold function, called a discriminant, is applied to the activation function's output to convert it into a binary classification ...