|
|
Neural Network Letter Recognition
Kluknavský, František ; Hradiš, Michal (referee) ; Šilhavá, Jana (advisor)
This work uses handwritten character recognition as a model problem for using multilayer perceptron, error backpropagation learning algorithm and finding their optimal parameters, hidden layer size, learning rate and length, ability to handle damaged data. Results were acquired by repeated simulation and testing the neural network using 52,152 English lowercase letters. Best results, smallest network and shortest learning time was at 60 neurons in the hidden layer and learning rate of 0.01. Bigger networks achieved the same ability to recognize unknown patterns and higher robustness at highly damaged data processing.
|
|
|
Graphical sign in the context of school and town
Novotná, Monika ; Kafková, Helena (advisor) ; Raudenský, Martin (referee)
NOVOTNÁ, Monika. Graphic sign in context of school and town. Prague, 2014.Thesis. Charles University in Prague, Faculty of Education, Department of Fine Arts. Supervisor Mgr. Helena Kafková, 90 pages. In the theoretical part I focus on a sign, the graphical sing. I sort out its forms. I present the sing as a nonverbal system of communication and a way of labelling objects. I search for the connections and contradictions of the existence of legal [traffic signs, orientation signs, signboards] and illegal graphic signs [graffiti and street art] in the public space. In the didactic part I suggest and realize the art projects where some technics of graffiti and street art are used. The artistic themes: SING - LETTER - PICTOGRAM - PUBLIC SPACE are processed at my art lessons at both primary and secondary schools. We discuss the meaning of these words in the public space. The part of the thesis is a research study among the pupils of a primary and secondary school which tries to point at either different or the same art perceptions by the pupils and their mental constructions of these. There I try to connect the street art and graffiti with the topics of private property, vandalism and personal attitudes versus the state and society. I offer to the children the question of personal freedom of an...
|
|
|
Neural Network Letter Recognition
Kluknavský, František ; Hradiš, Michal (referee) ; Šilhavá, Jana (advisor)
This work uses handwritten character recognition as a model problem for using multilayer perceptron, error backpropagation learning algorithm and finding their optimal parameters, hidden layer size, learning rate and length, ability to handle damaged data. Results were acquired by repeated simulation and testing the neural network using 52,152 English lowercase letters. Best results, smallest network and shortest learning time was at 60 neurons in the hidden layer and learning rate of 0.01. Bigger networks achieved the same ability to recognize unknown patterns and higher robustness at highly damaged data processing.
|
guest ::
