Artificial Neural Networks origins are in algorithms that try to mimic the brain and its neurons, back to the 40s of the past century.
They were widely used in 80s and early 90s but their popularity diminished in late 90s when they failed to keep up with the promises.
Their recent resurgence is due to the increased computation power (that allow bigger and deeper networks) and the availability of data (that allows proper training).
Nowadays are used in many cognitive applications (such as state-of-the-art for computer vision, speech recognition, automatic translation and many more) due to their self-learning ability and feature detection not possible with normal systems.
To describe the neural networks, we will begin by describing the simplest possible neural network, one which comprises a single “neuron”. Continue reading “Perceptron, an artificial neuron”