• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 2
  • 1
  • Tagged with
  • 3
  • 3
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Automatic Control of a Window Blind using EEG signals

Teljega, Marijana January 2018 (has links)
This thesis uses one of Brain Computer Interface (BCI) products, NeuroSky headset, to design a prototype model to control window blind by using headset’s single channel electrode. Seven volunteers performed eight different exercises while the signal from the headset was recorded. The dataset was analyzed, and exercises with strongest power spectral density (PSD) were chosen to continue to work with. Matlabs spectrogram function was used to divide the signal in time segments, which were 0.25 seconds. One segment from each of these eight exercises was taken to form different combinations which were later classified.The classification result, while using two of proposed exercises (tasks) was successful with 97.0% accuracy computed by Nearest Neighbor classifier. Still, we continued to investigate if we could use three or four thoughts to create three or four commands. The result presented lower classification accuracy when using either 3 or 4 command thoughts with performance accuracy of 92% and 76% respectively.Thus, two or three exercises can be used for constructing two or three different commands.
2

Metodologia de extração automática de características da mão para a estimação da idade óssea utilizando redes neurais artificiais no processo de decisão / Methodology of automatic extraction of hand characteristics for the estimation of the bone age using artificial neural nets in the decision process

Queiroz, Alini da Cruz 26 May 2006 (has links)
Este trabalho tem como objetivo principal apresentar uma metodologia para estimação da idade óssea baseada no método de Eklof & Ringertz utilizando redes neurais artificiais como classificador, com a finalidade de auxiliar o diagnóstico do radiologista e diminuir a dimensionalidade dos dados analisados pela rede neural, diminuindo a quantidade de centros de ossificação do método utilizado. A metodologia contém um processo automático de extração de características de imagens radiográficas da mão. Na etapa de classificação é utilizada a rede neural perceptron multicamadas, com o algoritmo de treinamento de Levenberg-Marquardt. As características extraídas da imagem são utilizadas como entrada para a rede neural, e os dados do Atlas de Eklof & Ringertz são utilizados como matriz de treinamento. Os resultados da etapa de classificação chegaram a uma taxa de 95% de acerto ao utilizar um centro de ossificação a menos que o método de Eklof & Ringertz simplificado / Grounded an Eklof & Ringertz’s method and using artificial neural networks as classifier, the main purpoise of this work is to present a methodology to reckon the bone age to the effect to help the radiologist’s diagnosis and to reduce the dimensionality of the data analyzed by neural network, reducing the quantity of the ossification’s centers of the used method. The methodology holds an automatic process to the hands radiographies image’s features. The multilayer perceptron neural network is used in the classification stage, with the Levemberg-Marquardt’s training algorithm. The taken image’s features are used as an input to the neural network, and Eklof & Ringertz’s Atlas data are used as training source. The results of the classification stage reached a rate of 95% of accuracy when applying the Eklof & Ringertz’s simplified method, excluding one of the ossification center
3

Metodologia de extração automática de características da mão para a estimação da idade óssea utilizando redes neurais artificiais no processo de decisão / Methodology of automatic extraction of hand characteristics for the estimation of the bone age using artificial neural nets in the decision process

Alini da Cruz Queiroz 26 May 2006 (has links)
Este trabalho tem como objetivo principal apresentar uma metodologia para estimação da idade óssea baseada no método de Eklof & Ringertz utilizando redes neurais artificiais como classificador, com a finalidade de auxiliar o diagnóstico do radiologista e diminuir a dimensionalidade dos dados analisados pela rede neural, diminuindo a quantidade de centros de ossificação do método utilizado. A metodologia contém um processo automático de extração de características de imagens radiográficas da mão. Na etapa de classificação é utilizada a rede neural perceptron multicamadas, com o algoritmo de treinamento de Levenberg-Marquardt. As características extraídas da imagem são utilizadas como entrada para a rede neural, e os dados do Atlas de Eklof & Ringertz são utilizados como matriz de treinamento. Os resultados da etapa de classificação chegaram a uma taxa de 95% de acerto ao utilizar um centro de ossificação a menos que o método de Eklof & Ringertz simplificado / Grounded an Eklof & Ringertz’s method and using artificial neural networks as classifier, the main purpoise of this work is to present a methodology to reckon the bone age to the effect to help the radiologist’s diagnosis and to reduce the dimensionality of the data analyzed by neural network, reducing the quantity of the ossification’s centers of the used method. The methodology holds an automatic process to the hands radiographies image’s features. The multilayer perceptron neural network is used in the classification stage, with the Levemberg-Marquardt’s training algorithm. The taken image’s features are used as an input to the neural network, and Eklof & Ringertz’s Atlas data are used as training source. The results of the classification stage reached a rate of 95% of accuracy when applying the Eklof & Ringertz’s simplified method, excluding one of the ossification center

Page generated in 0.0704 seconds