01Overview
In 1943 McCulloch and Pitts proposed an artificial neuron; in 1958 Rosenblatt built the perceptron: a weighted sum of inputs, a threshold, and weights corrected by mistakes. It could learn any linearly separable problem and was used for simple image recognition.
In 1969 Minsky and Papert published Perceptrons, proving that a single layer cannot solve non-linear problems such as XOR. The book is often misread as a refutation of neural networks in general, yet it did hasten connectionism’s first winter. Multi-layer networks and backpropagation had to wait until the 1980s.
02History
- 1943
The McCulloch–Pitts neuron
Neural activity described with logic gates.
- 1958
The perceptron
A learnable linear classifier.
- 1969
Perceptrons (the book)
It points out the limits of single-layer networks.
- 1986
Backpropagation
Multi-layer networks become trainable.
03Core concepts
Linear classification
Separating two classes with a hyperplane.
Weight learning
Adjusting weights from errors — the ancestor of gradient descent.
The XOR problem
The classic linearly inseparable counter-example.
04Hands-on
Do it once — it beats reading it ten times
Train a perceptron by hand on AND, OR and XOR; show why the third cannot converge
05Self-check
Say it in your own words first, then open the answer. Whatever you cannot explain is what to review next.
Can you explain "Linear classification"?
Separating two classes with a hyperplane.
Can you explain "Weight learning"?
Adjusting weights from errors — the ancestor of gradient descent.
Can you explain "The XOR problem"?
The classic linearly inseparable counter-example.
What problem did it solve, and why was there nothing better at the time?
Go back to the history section: how was this done before, and what did it cost?
If you had to build it today, how would you break it down?
Use the hands-on task above: write a plan of at most three steps before you start.
06Key people
Frank Rosenblatt
1928 – 1971
Inventor of the perceptron.
Marvin Minsky
1927 – 2016
Co-author of Perceptrons.
07Further reading
- Perceptrons
- Principles of Neurodynamics