Perceptron
Rosenblatt described a trainable recognition model.
Why it matters
It made learning from examples a concrete research programme.
The historical perceptron differs substantially from today’s deep networks.
Atlas record: https://www.aitimeline.dev/en/events/evt-0003/
Research · November 1958
Placed by the contemporary primary publication. The exact event date is not known; its documented interval appears below.
Event record
- Event date
- 1958
- Timeline date
- Primary publication date
- Verification
- Sources gathered automatically · September 11, 2026
- Lines
- ID
- evt-0003
Sources
- primary The Perceptron
Psychological Review · Published November 1958
- secondary Professor's perceptron paved the way for AI - 60 years too soon
Cornell Chronicle · Published September 25, 2019
Related events
Earlier
- Related McCulloch and Pitts neuron model 1943links to this record
The perceptron added a learning procedure to the formal neuron.
Later
- Related ADALINE and the least-mean-squares rule 1960links to this record
Both train a linear unit from examples; ADALINE uses a different weight-update rule.
- Builds on Mark I Perceptron: the perceptron in hardware 1960links to this record
The Mark I, by its engineers' article, was built to verify the mathematical predictions of Rosenblatt's perceptron.
- The Mark I Perceptron (John C. Hay, Charles W. Wightman), Research Trends (Cornell Aeronautical Laboratory) VIII(1), 1-4, Spring 1960 · Cornell Aeronautical Laboratory (original issue scanned in the Internet Archive)
- Related A recogniser that makes its own operators 1961links to this record
The paper mentions Rosenblatt's Perceptron: random machines, by the authors, identify poorly, and the only exception is Roberts's modification, which makes the Perceptron «essentially non-random»; the authors' program is set against both.
- Extends Principles of Neurodynamics 1962links to this record
The book develops the model first published in 1958.
- Principles of Neurodynamics: Perceptrons and the Theory of Brain Mechanisms · Cornell Aeronautical Laboratory
- Extends The perceptron convergence proof 1962links to this record
The proof concerns the perceptron learning algorithm itself.
- On Convergence Proofs for Perceptrons · Stanford Research Institute
- Builds on Glushkov: Introduction to Cybernetics 1964links to this record
The book's chapter on training pattern recognition sets out a theory of discrete alpha-perceptrons and cites Rosenblatt's perceptron.
- Introduction to Cybernetics, by V. M. Glushkov (FTD-TT-65-942/1+2, unedited rough draft translation) · Foreign Technology Division, Air Force Systems Command (Defense Technical Information Center AD0633642)
- Contrasts Perceptrons: the limits of a single layer 1969links to this record
The book formally analyses which problems a single-layer perceptron cannot solve.
- Builds on WISARD: a neural network made of memory chips 1984links to this record
Aleksander places WISARD in the line of neural modelling that includes Rosenblatt's perceptron, and calls it a single-layer neural net whose neurons are memory chips.
- Memory Networks for Practical Vision Systems: Design Calculations (I. Aleksander), chapter 11 of Artificial Vision for Robots, ed. I. Aleksander, 197-214, 1983: publisher's two-page preview · Springer (Kogan Page edition, preview of pages 197-198)
- Contrasts Mead names neuromorphic systems 1990links to this record
The perceptron took the idea of learning from the nervous system; Mead proposes taking its physics of computation: analog circuits that perform the operations neurons do with charge and current.
- Neuromorphic Electronic Systems (Carver Mead), Proceedings of the IEEE 78(10), 1629-1636, October 1990 · IEEE (copy in CaltechAUTHORS, California Institute of Technology)