The perceptron algorithm learns the weights of a linear binary classifier by processing training examples one at a time, adjusting the weight vector whenever the current classifier misclassifies an example, moving the decision boundary in the direction that would have classified that example correctly, and repeating over the training set until no more mistakes are made or a maximum number of passes is reached. Frank Rosenblatt introduced it in 1958 as a model of a simplified biological neuron and the learning rule for one of the earliest artificial neural network models. It is guaranteed to converge to a separating hyperplane when the training data is linearly separable, the perceptron convergence theorem.
Facts
Partially Attested
Time ComplexityAt most (R/gamma)^2 mistakes before convergence on linearly separable data 2 Source gives a mistake bound (Novikoff bound) for linearly separable data, not an explicit running time Time Complexity
Time Complexity (category) Connections
Associated With
In Field
Source Wikipedia: Perceptron
Sources
1. Perceptron Algorithm (Wikipedia)
Wikipedia infobox: time complexity linearQuote, Wikipedia infobox: time complexity linear
linear
2. Wikipedia: Perceptron
History
Rosenblatt described the details of the perceptron in a 1958 paper
Convergence of one perceptron on a linearly separable dataset
converges after making at most (R / γ)² mistakes
In Field: Machine Learning, Lead sentence
perceptron is an algorithm for supervised learning of binary classifiers.
In Group: Neural Network Training and Representation Algorithms, Lead
In machine learning, the perceptron is an algorithm for supervised learning of binary classifiers.
View the SourceReader Challenges (0)
No disputes yet. Spotted an error or a better source? Open the first one.
Sign in to dispute this or suggest a correction.