Artificial neural network (ANN) is a simplified machine learning model of brain-like systems. This model (along with its extension to deep neural networks) is believed to be the most powerful machine learning model we have at the moment. With the advancement of neuroscience, human is getting more knowledges of how our brain is working. The brain is believed to have connectionist architectures, parallel distributed-processing systems, and neuromorphic computation. Based on these knowledges, AI researchers designed ANN with a brain-style computation. But I need to mention that, although ANN research is continuously getting inspiration from neuroscience researches, and many ANN researchers pays intensive attention to neuroscience and its related subjects (including cognition science, psychology, etc.), most achievements in ANN were driven by the development of engineering methods.
Both NN and ANN use the neuron as the basic processing unit. A neuron or nerve cell is an electrically excitable cell that communicates with other cells via specialized connections called synapses. Synapse is the main component of nervous tissue in almost all animals. A typical neuron consists of a cell body (soma), dendrites, and a single axon. The soma is usually compact. The axon and dendrites are filaments that extrude from it. Dendrites is the signal receiver, and they typically branch profusely and extend a few hundred micrometers from the soma. The axon leaves the soma at a swelling called the axon hillock, and travels for as far as 1 meter in humans or more in other species. At the farthest tip of the axon's branches are synapses, where the neuron can transmit a signal to another cell [].
The neurons in the human brain are diverse and numerous. In fact, the human brain has about 86 billion neurons, according to Brazilian neuroscientist Suzana Herculano Hozel, who processed four brains into "brain soup", and then calculated the number of nucleis to determine the true number of neurons in the human brain [].
In ANN, neuron is characterized by an activity level, an output value, a set of input connections, a bias value, and a set of output connections. All of the aspects are represented mathematically by real numbers, so each connection has an associated weight, or synaptic strength, that determines the effect of the incoming input on the unit's activation level.
The first and simplest ANN neuron is perceptron. A perceptron is a neural network unit (an artificial neuron) that does certain computations to detect features or business intelligence in the input data.
Perceptron was introduced by Frank Rosenblatt in 1957. A Perceptron is an algorithm for supervised learning of binary classifiers. This algorithm enables neurons to learn and processes elements in the training set one at a time.
