1. Single-layer perceptron
history:
Background of the Perceptron
In 1958, Psychological Review published Cornell University researcher Frank Rosenblatt's "The Perceptron: A Probabilistic Model of Information Storage and Organization in the Brain ."
The paper formally introduced the concept of a perceptron — a linear classifier — that, according to Rosenblatt, has the potential to mimic the functioning of the human brain. (A year earlier, in 1957, Rosenblatt had written a report on Perceptrons for an internal research project at Cornell Aeronautical Laboratory )
Soon after, media outlets including The New York Times and The New Yorker published articles about Rosenblatt's work. Interestingly, some of Rosenblatt's predictions about the future of perceptrons turned out to be surprisingly accurate.
"It is predicted that later perceptrons will be able to recognize people and call out their names, and instantly translate speech in one language into speech or text in another." The
New York Times, "New Navy Equipment Learns Through Practice" 1958 July 7th.
In 1969, Marvin Minsky and Seymour Papert published The Perceptron—a historical work that changed the course of artificial intelligence research for more than a decade.
Minsky and Papert proved that a single perceptron—the grandparent of the computational units that make up modern neural networks—cannot learn the exclusive-or (aka XOR) function. Leading to the cold winter of more than ten years of artificial intelligence, the multi-layer perceptron can solve the XOR (aka XOR) function in about fifteen years.
2. Multilayer Perceptron
3 Scratch implementation of multilayer perceptron
4. Introduction to Implementation of Multilayer Perceptron
refer to
https://www.bilibili.com/video/BV1hh411U7gn?p=1
https://alan.do/minskys-and-or-theorem-a-single-perceptron-s-limitations-490c63a02e9f