ALGORITHMS

Algorithms

Abstract

This research paper presents the concepts of machine learning through supervised algorithms. There are several suggested algorithms and, we may discuss several algorithms. This paper is focusing on the machine learning through the parameters of the algorithms suggested.

Algorithms

Direction of artificial intelligence associated with the development and construction of analytical models that are able to automatically detect the hidden and previously unknown patterns as well as to independently acquire the properties necessary for the implementation of these laws.

Bayesian Classifier

To define the Bayesian classifier, a cost measure is required, which assigns to each possible classification of costs. The Bayes classifier is precisely the one classifier that minimizes the costs incurred by all classifications (Yang et al, 2003, pp. 245-258). The cost measure is sometimes called risk function, then said to the Bayes classifier that minimize the risk of wrong decision and defined by the minimum-risk criterion. It is a primitive cost measure that causes only when mistakes cost, minimizes the Bayes classifier, the probability of a wrong decision. Individual believes that he is defined by the maximum a posteriori criterion. Both forms assume that the probability that a point of the feature space belongs to a particular class, is known to each class by a probability density is described (Yang et al, 2003, pp. 245-258). In reality, these density functions are not known, and they must be estimated. This is supposed to be behind each class one type of probability distribution - usually a normal distribution, and tried using the available data to estimate the parameters. The Bayes classifier is used to assess other classifiers such as; it artificially creates some classes and their probability densities generated with this model, and a random sample can be divided into the other classifier, and the objects in this sample classes. The result is compared with the classification that would have made the Bayesian classifier. Since the Bayesian classifier is optimal in this case, and we obtain an estimate of how close the other classifier is the optimum. Simultaneously, the Bayes classifier is a lower bound for the error probability of all other classifiers in this scenario, better than the optimal Bayes classifier cannot be like this. In the course of machine learning, different types of classifiers have been developed, always with the aim to achieve the highest degree of accuracy and efficiency, each with its advantages and disadvantages. However; they share common characteristics and, the construction of the majority of them consists of two main steps. First, the definition of a function which associates to a document a value between 0 and 1 representing the degree of membership in the category.

The importance of choosing the threshold has been discussed in more depth in [Yan01], which has among others the following strategies: Assign each document n categories for which the classifier will have given the highest scores Associated with each category n best documents, that is to say, documents for which the classifier will be given the highest scores for the category in question Select ...