Information / AI & Robotics

Perceptron & Artificial Neurons

Rosenblatt’s perceptron learns from examples instead of being programmed: the ancestor of all neural networks.

After the mathematical neuron model of McCulloch and Pitts (1943), Frank Rosenblatt built the perceptron at Cornell in 1958, which learned to recognize letters. Minsky and Papert showed in 1969 that simple perceptrons cannot solve some tasks (for example XOR), which stalled funding for neural networks for a decade. The disillusionment fed the first “AI winter” of the 1970s.