چكيده به لاتين
Decoding of brain signals can be one of the most fascinating human research to find facts, however small, in the face of this great creation. The relationship between environmental findings such as seeing, hearing and touching and useful information in cognitive science and brain signals can be acquired at the disposal of mankind. In this thesis, decoding of brain signals based on the number of images have been viewed by those surveyed. Visual stimuli, including images of handwritten numbers one to three , which is a database of handwritten numbers-have been randomly selected standard. The images were randomly divided into seven volunteers who record the brain signals of them are shown. Wavelet transform method for feature extraction and feature selection and feature ranking methods have been used to reduce post. To classify these signals such as neural networks MLP artificial neural networks and neural networks radial basis functions are used. As well as the separator multi-class support vector machine classifier is used as another option. By applying different amounts of neurons and layers to determine the final structure of neural network classifiers, these values were determined. Classification results show that the neural network classifiers, radial basis functions neural network with 55 percent ttest Feature selection , and also support vector machine with 65 percent among classifiers had maximum efficiency in between classifiers
Keywords: Classification of neural networks, wavelet transform, feature extraction
