SDSU Department of Computer Science
  
Home Contact Us Undergraduate Student Info Graduate Student Info Faculty & Staff News & Events
News Calendar Problem of the Fortnight

April 2008 Calendar

CS Masters' Thesis Defense

Title: Recognition Of Eeg Epileptic Signals Using Artificial Immune System
Speaker: Marios Sofocleous
Date: Friday, April 25, 2008
Time: 1:00 p.m.
Location: GMCS 328
Thesis advisor: Dr Roman Swiniarski

Abstract:
This thesis presents an application of wavelets methods (DWT), Principal Component Analysis (PCA), and Artificial Immune Systems (AIS) in seizure recognition systems. More emphasis has been given in the classifier design for two reasons. First, because immune inspired algorithms are a fairly new area, and second, because the human immune system demonstrates features such as distributed control, parallel processing and adaptation, which could be useful metaphors in implementing machine learning algorithms. The Artificial Immune Recognition System (AIRS), is a supervised learning AIS that has shown significant success in a broad range of classification problems.

Most importantly, this research wants to demonstrate the great generalization ability of AIS applying on EEG signals in terms of data needed for classification and features per pattern. This is attributed to the fact that the classifier is able to produce a pool of vectors, which are used for classification using the K-nn method, and resembles the immune memory in the natural immune system generated after immunization. These cells not only are trained to recognize known pathogens (training data) but also are generalized well to recognize unknown but similar pathogens (test data). The classifier was able to generalize well by learning using only subsets of vectors randomly selected from the whole database and using only 5 features per pattern.
webmaster@cs.sdsu.edu page counter College of Sciences