neural network

A neural network is a computer program with a logical structure and strategy inspired by a brain's structure and operation, of neurons interconnected by synapses. The term may also refer to specialized computer hardware that carries out the same such computations (essentially, such a computer program implemented as hardware), that might also be physically structured as neurons and synapses. These can be used for classification applications such as recognizing pictures, and also can be made to program themselves as a kind of machine learning (ML). Their learning and classification may mimic biological perception, motor control, learning, and thought, but they aren't restricted to pure mimicry and likely their operation differs considerably from brain operation of a human carrying out the same task. With the development of such techniques and with increasing computational capacity, the method has become very effective for a number of types of applications, including recognizing patterns such as patterns within images, a common use in astronomy.

The term deep learning is often used for machine learning using neural networks, but sometimes it is used more specifically, such as for neural networks with some minimal number of layers: neural networks are generally organized in a series of layers of (artificial) neurons (programs that process data much like a nervous system's neuron does): input data to a neuron in a layer after the first is from multiple neurons in the first layer, and its output is directed to multiple neurons in the following layer. As computing progress has allowed larger (and deeper) such networks to be implemented practically.

The term neural network is likely sometimes used to describe the structure of actual brains.