Comprehensive Analysis of Extreme Learning Machine and Continuous Genetic Algorithm for Robust Classification of Epilepsy from EEG Signals - Softcover

Prabhakar, Sunil Kumar

Softcover

ISBN 10: 3960670990 ISBN 13: 9783960670995

Verlag: Anchor Academic Publishing, 2017

Inhaltsangabe

Epilepsy is a common and diverse set of chronic neurological disorders characterized by seizures. It is a paroxysmal behavioral spell generally caused by an excessive disorderly discharge of cortical nerve cells of the brain. Epilepsy is marked by the term "epileptic seizures". Epileptic seizures result from abnormal, excessive or hyper-synchronous neuronal activity in the brain. About 50 million people worldwide have epilepsy, and nearly 80% of epilepsy occurs in developing countries. The most common way to interfere with epilepsy is to analyse the EEG (electroencephalogram) signal which is a non-invasive, multi channel recording of the brain's electrical activity. It is also essential to classify the risk levels of epilepsy so that the diagnosis can be made easier.
This study investigates the possibility of Extreme Learning Machine (ELM) and Continuous GA as a post classifier for detecting and classifying epilepsy of various risk levels from the EEG signals. Singular Value Decomposition (SVD), Principal Component Analysis (PCA) and Independent Component Analysis (ICA) are used for dimensionality reduction.

Über diesen Titel

Auszug. © Genehmigter Nachdruck. Alle Rechte vorbehalten.

Text Sample:
Chapter 4. Continuous Genetic Algorithm as a Post Classifier for Epilepsy Risk Level Classification:
Genetic algorithm is based on biological concept of generation of the population, a rapid growing area of Artificial intelligence. GA's are inspired by Darwin's theory about Evolution. According to the Darwin "Survival of the fittest". Genetic Algorithm is based on the biological concept of population generation.
A typical genetic algorithm requires:
i. A genetic representation of the solution domain.
ii. A fitness function to evaluate the solution domain.
Once the genetic representation and the fitness function are defined, a GA proceeds to initialize a population of solutions and then to improve it through repetitive application of the mutation, crossover, inversion and selection operators. GA has blossomed rapidly due to the easy availability of low cost but fast speed small computers. The complex and conflicting problems that required simultaneous solutions, which in past were considered deadlocked problems, can now be obtained with GA. However, the GA is not considered a mathematically guided algorithm. The optima obtained are evolved from generation to generation without stringent mathematical formulation such as the traditional gradient-type of optimizing procedure. In fact, GA is much different in that context. It is merely a stochastic, discrete event and a non linear process. The obtained optima are an end product containing the best elements of previous generations where the attributes of a stronger individual tend to be carried forward into the following generation. The rule of the game is "survival of the fittest will win" [21].
Most values in the real world are neither integers nor in encoded form. They appear as real values and the advancement in computing has enabled us in realizing these values to a high precision. Hence there is requirement of an algorithm, which can be used in manipulating these real valued data. In a similar sense, the output of the fuzzy system is also a real number in the '0 to 20' scale with a precision of 4 digits after decimal point. Instead of encoding the fuzzy output in the form of strings and converting them to bit representation, we can directly use the values in a continuous parameter genetic algorithm (CGA) to obtain more precise outputs. The advantages of using a CGA are:
a. Reduction in memory size occupied.
b. Reduction in quantization error.
c. Increase in precision.
d. Speedup of computation.
The iterations in a continuous parameter genetic algorithm follow the footsteps of its predecessor. The difference in the continuous parameter genetic algorithm occurs in the computation of the fitness function and the crossover and mutation operators. The following section elucidates as to how the operators in a continuous parameter genetic algorithm differ from a conventional binary genetic algorithm.
4.1 Functions and Operators:
Fitness function:
The fitness function is calculated on a 0 to 1 scale. As the output risk level can have a maximum value of 20, dividing by 20 normalizes each gene and the average of all six genes is taken as the fitness of the chromosome. This fitness value of the chromosome is used in selecting the chromosome for the next level of iteration after sorting.

Details zur Ausgabe

• Verlag: Anchor Academic Publishing
• Erscheinungsdatum: 2017
• Sprache: Englisch
• ISBN 10: 3960670990
• ISBN 13: 9783960670995
• Einband: Taschenbuch
• Anzahl der Seiten: 36
• Kontakt zum Hersteller: preigu GmbH & Co. KG; mail@preigu.de; ; Lengericher Landstr. 19; Osnabrück; Niedersachsen; 49078; Deutschland