论文信息 - The Use of Feed Forward Neural Network for Recognizing Characters of Dactyl Alphabet

The Use of Feed Forward Neural Network for Recognizing Characters of Dactyl Alphabet

This paper deals with recognition in dependency of count of objects in the training data set. For the capture the fingers position is used the sensory glove 5DT Data Glove Ultra.We describe the computer software "Dactyl Teacher" for recognition characters of Dactyl alphabet. This software contains a feed forward neural network, which is used for character recognition. In experimental part we examined count of iterations to train the neural network in influence of count of hidden layers and of count of neurons in these hidden layers. We choose the sufficient combination of these parameters for our feed forward neural network. Next we examined the success of recognition in dependence of count of training objects in training data set.

Roman Záluský | Daniela Durackova | Emil Raschman | Mário Krajmer

[1] Shahrul Azman Noah,et al. Performance Comparison of Multi-layer Perceptron (Back Propagation, Delta Rule and Perceptron) algorithms in Neural Networks , 2009, 2009 IEEE International Advance Computing Conference.