One class classifiers based on artificial immune systems
Μονοταξικοί ταξινομητές βασισμένοι σε τεχνητά ανοσοποιητικά συστήματα
KeywordsOne class classification ; One class classifiers ; Artificial immune systems ; A.I. ; Real-valued negative selection ; Negative selection ; V-Detector
In this Master’s Thesis we are going to talk about some basic principles of machine learning and we are going to give the definitions about some basic concepts. Then we will focus our interest to the one class classification problem. Furthermore, we will talk about some biological processes, their use in computer science and what is an artificial immune system. Then, we will try to describe how could, an artificial immune system, be used in the problem of one class classification. We will mention the Negative Selection, Real-Valued Negative Selection algorithms and we will focus our attention on the V-Detector algorithm, with whom we are going to perform our experiments later. For our experiments we will have to generate our own artificial data according to a specific geometric structure. This procedure is analyzed in section 4. Finally, we are going to present the results of our experiments and we will try to evaluate the algorithm and draw some conclusions.