Distributed Genetic Algorithm with Bi-Coded Chromosomes and a New Evaluation Function for Features Selection

Dynamic Adaboost learning with feature selection based on parallel genetic algorithm for image annotation

“…The fitter chromosomes have higher probability to be selected for GA operation. We involve two kinds of GA: binary-coded chromosome GA and bi-coded one [4,50,51]. Applying GA, we need to take into account four basic factors: chromosome design, initial population, fitness function and genetic operator design.…”

Feature selection based-on genetic algorithm for image annotation

“…Chromosomes are selected for reproduction by evaluating the fitness value. The fitter chromosomes have higher probability to be selected for GA operation [16,17,23].…”

“…In [16], Hamdani et al proposed a bi-coded chromosome genetic algorithm to simultaneously select feature subset and the importance rate of each feature. They combined the k-NN recognition rate and the size of the selected feature subset as fitness function.…”

“…Three schemes are considered: weight optimization with a real coded chromosome GA, the selection for optimal feature descriptor subset with a binary one, the simultaneity of weight optimization and the selection for optimal feature descriptor subset with a bi-coded chromosome GA motivated by [16]. As we just want to investigate the methodology in feature selection, the fitness function takes into consideration k-nearest neighbor (k-NN) classification accuracy instead of SVM one, which is one of the best classification algorithm [17].…”

An Iterative Method for Deciding SVM and Single Layer Neural Network Structures

“…In SVM and neural based classifiers, this problem is widely studied especially in pattern recognition when training data points aren't sufficiently high or when they are of a very high dimensions [35,36].…”