Perceptron
lastmod 2024-05-08created 2015-11-0317 edits
The perceptron is a type of artificial neural network, invented in 1957 by Frank Rosenblatt at the Cornell Aeronautical Lab. It can also be seen as the simplest form of a feedforward network - a linear classifier -.
A perceptron works as follows. The sum of the products of each node's weight and input is evaluated by an activation function; if that value is greater than a threshold (usually 0), the neuron is activated and outputs 1. If the neuron is not activated, it outputs -1.
In their book "Perceptrons", Marvin Minsky and Seymour Papert showed that a single-layer perceptron cannot perform the XOR operation(because it can only have a discriminant expressed as a single straight line), but a multilayer perceptron can.
What is perceptron
A perceptron takes several binary inputs x1, x2 ... and produces a single binary output: ArtificialNeuronModel_english.png
Rosenblatt proposed a formula for computing the output. He introduced weights such as w1, w2.., which express the relative importance of each input. The neuron's output, 0 or 1, is determined by whether the sum of the products of each weight and input is greater or less than a given threshold value. Like other parameters such as the wieght, the threshold is a real number. Expressed as a formula:
\[ \begin{eqnarray} \mbox{output} & = & \left\{ \begin{array}{ll} 0 & \mbox{if } \sum_j w_j x_j \leq \mbox{ threshold} \\ 1 & \mbox{if } \sum_j w_j x_j > \mbox{ threshold} \end{array} \right. \tag{1} \end{eqnarray} \]
That's all there is to how a perceptron works!
Example
Multilayer perceptron
For details, go to that page.
See also
Favorite site
- Wikipedia (en) Explanation of Perceptron
- [Recommended] Machine Learning Basics 1
- Coursera Neural Networks for Machine Learning Week2 - Perceptron
- Single Layer Perceptron
- Deep learning_Neural Network_Perceptron 2
- Introduction to machine learning models (6) - Multilayer Perceptron
References
-
Whydsp.org_-_Machine_Learning_Basics.pdf ↩