The perceptron algorithm was invented around 1959 by Cornell psychologist Frank Rosenblatt (1928-1971), a pioneer of the field artificial intelligence. As we will see in this module, the vanilla perceptron is basically a single neuron, and multilayer perceptron is the foundation of today’s deep learning. Therefore, Rosenblatt is sometimes called the father of deep learning, and his work on perceptron was ~60 years ahead of time.
However, the research on perceptron came to a halt around 1969 due to inability to solve inseparable cases, and it was not until 30 years later in 1999 that perceptron was revived, with the new form of large-margin classifiers, voted and averaged perceptron (Freund and Shapiro, 1999). Starting from 2012, the multilayer version of the perceptron is widely used in the deep learning revolution that has deeply impacted our daily lives.