Supervised classification algorithms

Supervised classification algorithms. The algorithms of supervised classification used in problems which are known a priori the number of classes and [[Pattern Recognition | patterns] representatives of each class. Basically it consists in that, to automatically classify a new sample, the information that can be extracted from a set of available objects divided into classes and the decision of a classification rule or classifier are taken into account.

Summary

[ hide ]

  • 1 Concept
  • 2 Purpose of supervised classification
  • 3 Approach based on neighborhood criteria
  • 4 Example of classifiers based on neighborhood criteria
  • 5 Source

Concept

Algorithms dedicated to the problem of supervised classification usually operate on the information supplied by a set of training samples, patterns, examples or prototypes that are assumed to be representatives of the classes, and they have a correct class label. This set of correctly labeled prototypes is called a training set ( TS ), and it is the knowledge used to classify new samples.

Purpose of supervised classification

These algorithms are intended to determine which class, which is already known, to which a new sample should belong, taking into account the information that can be extracted from the training set.

Approach based on neighborhood criteria

Among the supervised classification algorithms are those that use neighborhood criteria . In general, any supervised classification problem addressed with a neighborhood criteria approach can be characterized as follows:

Sea E space representation of a given problem which have M classes, so as to form a partition of E . There are also N prototypes 1 , 2 , …, N (or labeled samples) belonging to space E , which are taken as a training set, which is represented by

TS = { X , Ω } = {( 1 , ω 1 ), ( 2 , ω 2 ), ( N , ω N )}

The problem will be that, given a new sample x ε E , statistically independent of the set { X , Ω } that can be contained in any of the M classes, determine to which class of the space it belongs. This procedure is known as the Classification Rule or Classifier and is represented as:

δ: E → Ω , δ ( x ) = ω i for some i = 1, …, M

That is, the classifier determines which class of the representation space the new sample x belongs to, taking into account the distance determined by a metric between x and the prototypes of the training set.

Example of classifiers based on neighborhood criteria

Among the supervised classifiers based on neighborhood criteria are the Nearest Neighbor Rule ( NN rule ) and the K Nearest Neighbor Rule ( k-NN rule )

 

by Abdullah Sam
I’m a teacher, researcher and writer. I write about study subjects to improve the learning of college and university students. I write top Quality study notes Mostly, Tech, Games, Education, And Solutions/Tips and Tricks. I am a person who helps students to acquire knowledge, competence or virtue.

Leave a Comment