|
|
Self-organization and artificial neural networks for knowledge extraction
Aharkava, Larysa ; Mrázová, Iveta (advisor) ; Iša, Jiří (referee)
Neural networks are widely used for nancial time series prediction. However, the future values' prediction has its drawbacks and often cannot be converted to the e ffective and pro table trading system. In that thesis I will describe several di erent types of neural networks. Then, I will propose and evaluate on real series data two di erent approaches based on Kohonen's self-organizing maps and back propagation networks of how to use those networks for creating successful and pro table trading models. Also, I will give a general overview of the Forex market (Foreign exchange market) and neural networks' usage within that market.
|
|
|
Artificial neural networks for pattern recognition
Kukačka, Marek ; Mrázová, Iveta (advisor) ; Božovský, Petr (referee)
This work describes the advantages and disadvantages of using neural networks for pattern recognition. Several neural network models are described and their use for pattern recognition is demonstrated. Standard multi-layered perceptron model is compared to a more sophisticated convolutional network model. A new network model is introduced, which is inspired by the convolutional networks and aimed at rectifying some of their shortcomings. The work describes results of tests performed with the described network model on the problem of recognizing hand-written digits.
|
|
|
Knowledge Extraction from Data
Kozák, Vladislav ; Mrázová, Iveta (advisor) ; Petříčková, Zuzana (referee)
The task of this master thesis is to describe the overall process of data mining and algorithms used for data preparation and data modelling. The qualities of these algorithms are compared and the results are well-founded with repeatable tests. Knowledge gained by this research is applied to 2 real data based tasks. Master thesis includes development of own data mining application. The stress was laid on robustness, intuitive GUI as well as wide spectrum of data mining algorithms implemented.
|
|
|
Deep neural networks and their implementation
Vojt, Ján ; Mrázová, Iveta (advisor) ; Božovský, Petr (referee)
Deep neural networks represent an effective and universal model capable of solving a wide variety of tasks. This thesis is focused on three different types of deep neural networks - the multilayer perceptron, the convolutional neural network, and the deep belief network. All of the discussed network models are implemented on parallel hardware, and thoroughly tested for various choices of the network architecture and its parameters. The implemented system is accompanied by a detailed documentation of the architectural decisions and proposed optimizations. The efficiency of the implemented framework is confirmed by the results of the performed tests. A significant part of this thesis represents also additional testing of other existing frameworks which support deep neural networks. This comparison indicates superior performance to the tested rival frameworks of multilayer perceptrons and convolutional neural networks. The deep belief network implementation performs slightly better for RBM layers with up to 1000 hidden neurons, but has a noticeably inferior performance for more robust RBM layers when compared to the tested rival framework. Powered by TCPDF (www.tcpdf.org)
|
|
|
Convolutional neural networks and their implementation
Schmid, Martin ; Mrázová, Iveta (advisor) ; Petříčková, Zuzana (referee)
Bachelor thesis describes using convolutional neural networks for recognizing symbols from images. First describes this model and shows it's implementation. Then this implementation is used for sample application. First, model of neural networks is described, then learning of this model (including backpropagation algorithm). Finally, convolutional neural networks are presented with it's advantages for symbol recognition. Then some existing implementations of neural networks are analyzed, including speed comparison. None of these implementations support convolutional networks, so this model is added to one of them. Then this extension and it's interface (how to use it) is presented. To show features of this model and to prove functionality of the implementation, sample application is created. This application is available on the web site and runnable using only a web browser. Keywords: Convolutional neural networks, OCR, Encog 7
|
|
|
Uživatelsky orientovaný jazyk pro řešení úloh DZD
Kováč, Michal ; Rauch, Jan (advisor) ; Mrázová, Iveta (referee)
The thesis discusses a new visual functional programming language and its use for data mining. The language is called Ferda and forms an integral part of the Ferda system, which is an application that has been created for data mining with the GUHA method. Functions of the language are represented by boxes. The source code is written as a connection of boxes; source files are project files of the Ferda system. The thesis describes the status of the Ferda system before this work from the point of view of the visual programming and describes possible enhancements to source files, then it presents a new basic set of boxes for the Ferda language and proposes other possible extensions of the language. Some of these proposals have been implemented as a part of this thesis. The last part includes examples of the use of the new language for data mining. One of these has also been implemented as a part of this thesis.
|
|
|
Feed-forward neural networks and their application in data mining
Civín, Lukáš ; Mrázová, Iveta (advisor) ; Štanclová, Jana (referee)
The goal of data mining is to solve various problems dealing with knowledge extraction from huge amounts of real-world data, the quality of which might be disputable. Neural networks can help with the solution due to their generalization capabilities. While working on data mining projects, we have essentially the following two objectives in real-world applications of feed-forward neural networks. To obtain applicable results, it is crucial to provide the networks with well-prepared data. However, it is equally important to choose the right training strategy for the networks themselves - including network architecture, parameter settings or the training algorithm. One of the most important ideas behind these steps is namely to prevent "over-training". The final network should recall unknown examples as well as possible. There are plenty of techniques with different approaches to the solution. It is possible to modify the data, these comprises modifying the range of the data or its dimension, adding noise to the data, etc. Yet another way is the modification of the neural network by structural learning with forgetting, weight decay or early stopping. These techniques are analyzes both theoretically and experimentally in this thesis. With regard to the results achieved in a number of experimental tests we have...
|
|
|
Convolutional neural networks and their application in object detection
Hrinčár, Matej ; Mrázová, Iveta (advisor) ; Pešková, Klára (referee)
1 Title: Convolutional neural networks and their application in object detection Author: Matej Hrinčár Department: Department of Theoretical Computer Science and Mathematical Logic Supervisor: doc. RNDr. Iveta Mrázová, CSc. Supervisor's e-mail address: Iveta.Mrazova@mff.cuni.cz Abstract: Nowadays, it has become popular to enhance live sport streams with an augmented reality like adding various statistics over the hockey players. To do so, players must be automatically detected first. This thesis deals with such a challenging task. Our aim is to deliver not only a sufficient accuracy but also a speed because we should be able to make the detection in real time. We use one of the newer model of neural network which is a convolutional network. This model is suitable for proces- sing image data a can use input image without any preprocessing whatsoever. After our detailed analysis we choose this model as a detector for hockey players. We have tested several different architectures of the networks which we then compared and choose the one which is not only accurate but also fast enough. We have also tested the robustness of the network with noisy patterns. Finally we assigned detected pla- yers to their corresponding teams utilizing K-mean algorithm using the information about their jersey color. Keywords:...
|
|
|
Reconstruction of the Hradisko Small Water Reservoir in the Radslavice Cadastral Area
Mrázová, Iva ; Kostelecký, Jiří (referee) ; Paseka, Stanislav (advisor)
This diploma thesis, called „Reconstruction of the small water reservoir Hradisko in the cadastral area of Radslavice“, focuses on the complex process of a reconstruction of the small water reservoir Hradisko. This work follows up on the bachelor’s thesis, in which the current state of the Hradisko reservoir was described in detail. Based on the survey of the dam and the flood using a GPS device, detailed project documentation was prepared. Within the solution for the reconstruction of Hradisko reservoir, a repair of the dam and the increase of the crown of the dam is proposed, new functional objects are dimensioned, the bottom is cleared of mud, including modifications in the flood, and other necessary steps for proper functioning of the crumbling no longer compliant reservoir are described. Functional objects are processed for two variants of the solution. The first option consists of the design of a bottom outlet and a safety spillway, and the second alternative involves the design of a combined functional block. In the end, the total costs of both options are quantified and compared on the basis of an indicative item budget.
|
|
|
Neural network architectures for mobile devices
Georgiev, Georgi Stoyanov ; Mrázová, Iveta (advisor) ; Božovský, Petr (referee)
Designing effective methods for image classification and real-time object detection is one of the most well-known problems of the present. A series of convolutional neural networks has been designed in order to solve these tasks. Neural networks created spe- cifically for mobile devices are among the fastest ones. In this work we focus primarily on the MobileNetV2 and EfficientNetB0 models. We present their structure and compare them with one another. We research several algorithms designed to automatically build new neural network models as well. An essential part of the convolutional network design process is the optimization of their structure. We outline sensitivity analysis methods which help us observe how network inputs influence its outputs, and pruning methods designed to remove redundant neurons. In the end we demonstrate an example usage of the EfficientNetB0 model in a mobile appliaction created to classify cars. 1
|
guest ::
RSS feed.