Who created the perceptron, an early artificial neural network, in 1957?

Frank Rosenblatt created the perceptron, an early artificial neural network, in 1957.

Rosenblatt was an American psychologist working at the Cornell Aeronautical Laboratory. His perceptron was a computational model inspired by simplified descriptions of biological neurons. It accepted input signals, applied adjustable weights, and produced an output. During training, the weights could be changed when the system misclassified an example.

The perceptron became an important foundation for later neural-network research. The Mark I Perceptron was implemented in hardware and demonstrated publicly in 1958. A single-layer perceptron can learn certain linearly separable classification problems, but it cannot represent every logical relationship, including exclusive OR without additional structure. Criticism of these limitations contributed to reduced interest in neural networks during part of the 1970s, while later multilayer methods and improved computing eventually revived the field.