Artificial Neural Networks, more commonly referred to as neural networks, are networks of many simple processors that have a small amount of local memory. These processors, otherwise known as "units", are connected through numerous communication channels that carry encoded numeric data. The units operate only on their local data and on the inputs they receive via the connections. Historically, much of the inspiration for neural networks came from the desire to produce artificial systems capable of sophisticated, perhaps even intelligent, computations similar to those of the human brain.

Most neural networks have some sort of "training" rule whereby the weights of connections are adjusted on the basis of data. This allows neural networks to "learn" from examples and exhibit the capacity for generalization beyond the training data. Consequently, neural networks can be extremely useful for pattern recognition, in that they can be trained to detect complex relationships between inputs even when the statistical distributions of those inputs is unknown.