Frank Rosenblatt

Built the perceptron, the first trainable artificial neuron.

Biography

An American psychologist who developed the perceptron, a device that computes a weighted sum of its inputs and fires if a threshold is exceeded, together with a rule for adjusting the weights from labelled examples. It was implemented in hardware as well as simulated.

Key contributions

• The perceptron model and its learning rule (1957).
• Demonstrated that a classifier could be learned from examples rather than hand-designed.

Impact

The perceptron is the direct ancestor of the artificial neuron used in every modern network. Its documented inability to represent certain functions with a single layer is what motivated multilayer architectures.