|
|
Self-organization and morphing
Lessner, Daniel ; Mrázová, Iveta (advisor) ; Mráz, František (referee)
Morphing is a well-known visual effect. It is based on a uent transition of one image into another one, metamorphosis of a movie character into a bear is a possible example. Realization of such an effect requires accurate, concentrated and expensive effort of an animator. Development of tools and methods for problem solving comparable with human intelect is the subject of arti tial intelligence. Systems working with no human adjustments are often based on self-organisation. Self-organisation of a system is the appearance of complex behavior that isolated parts of the system couldn't reach. This thesis examines possibilities of application of self-organisational methods of artifi tial intelligence in morphing with the goal of reduction of the human assistence. The thesis includes information about some drafted techniques and results of experiments with the most successful technique. Experiments imply that it is possible to reach good results without human assistance if certain conditions are met.
|
|
|
The Implementation of an Artificial Intelligence in a Strategy Game Simulator
Fürbach, Radek ; Mrázová, Iveta (advisor) ; Chrpa, Lukáš (referee)
The aim of this thesis is a comparison of a few selected methods of an artificial intelligence in a specified strategy game. The thesis contains three parts. The first part specifies a model of the strategy game, whereat are simulated some experiments. It defines objects that occur in the game, relation among them, and used algorithms. The second part specifies of the artificial intelligence that is used in the strategy game. It explains the genetic algorithm and shows a few methods of so called selection, crossing, and mutation. It describes some basic artificial neural networks and their architectures. The last part describes several algorithms of the artificial intelligence using theory from the second part. It compares their efficiency on the simulated experiments.
|
|
|
Artificial Neural Networks and Their Usage For Knowledge Extraction
Petříčková, Zuzana ; Mrázová, Iveta (advisor) ; Procházka, Aleš (referee) ; Andrejková, Gabriela (referee)
Title: Artificial Neural Networks and Their Usage For Knowledge Extraction Author: RNDr. Zuzana Petříčková Department: Department of Theoretical Computer Science and Mathema- tical Logic Supervisor: doc. RNDr. Iveta Mrázová, CSc., Department of Theoretical Computer Science and Mathematical Logic Abstract: The model of multi/layered feed/forward neural networks is well known for its ability to generalize well and to find complex non/linear dependencies in the data. On the other hand, it tends to create complex internal structures, especially for large data sets. Efficient solutions to demanding tasks currently dealt with require fast training, adequate generalization and a transparent and simple network structure. In this thesis, we propose a general framework for training of BP/networks. It is based on the fast and robust scaled conjugate gradient technique. This classical training algorithm is enhanced with analytical or approximative sensitivity inhibition during training and enforcement of a transparent in- ternal knowledge representation. Redundant hidden and input neurons are pruned based on internal representation and sensitivity analysis. The performance of the developed framework has been tested on various types of data with promising results. The framework provides a fast training algorithm,...
|
|
|
RBF-networks with a dynamic architecture
Jakubík, Miroslav ; Mrázová, Iveta (advisor) ; Kukačka, Marek (referee)
In this master thesis I recapitulated several methods for data clustering. Two well known clustering algorithms, concretely K-means algorithm and Fuzzy C-means (FCM) algorithm, were described in the submitted work. I presented several methods, which could help estimate the optimal number of clusters. Further, I described Kohonen maps and two models of Kohonen's maps with dynamically changing structure, namely Kohonen map with growing grid and the model of growing neural gas. At last I described quite new model of radial basis function neural networks. I presented several learning algorithms for this model of neural networks, RAN, RANKEF, MRAN, EMRAN and GAP. In the end of this work I made some clustering experiments with real data. This data describes the international trade among states of the whole world.
|
|
|
Decision Trees and Knowledge Extraction
Vitinger, Jiří ; Mrázová, Iveta (advisor) ; Jiroutek, Pavel (referee)
The goal of data mining is to extract knowledge, dependencies and rules from data sets. Many complex methods were developed to solve it. This thesis presents some of the most important methods, which include the decision trees with algorithms ID3, C4.5 and CART, neural networks like multilayer neural networks with the backpropagation algorithm, RBF networks, Kohonens maps and some modifications of LVQ method. There are also described some clustering methods like hierarchical clustering, QT clustering, kmeans method and its fuzzy modification. The work also includes data pre-processing techniques, which are very important in order to obtain better results of data mining process. Experimental part of the work compares the presented methods by means of the results of many tests on real-world data sets. The results can be used as a guide to choose an appropriate method and its parameters for some given data set. In this work there is presented author's implementation of the decision trees C4.5 and CART in C#. In the application it is possible to watch details of algorithms work. The application provides an API enabling an implementation of new algorithms.
|
|
|
Multi-layered neural networks and visualization of their structure
Drobný, Michal ; Mrázová, Iveta (advisor) ; Kukačka, Marek (referee)
The model of multi-layered neural networks of the back-propagation type is well-known for their universal approximation capability and even the standard back-propagation training algorithm used for their adjustment often provides results applicable to real-world problems. The present study deals with the issue of the multi-layered neural networks. It describes selected variants of training algorithms, mainly the standard back-propagation training algorithm and the scaled conjugate gradients algorithm, which ranks among the extremely fast second-order algorithms. One of the parts of the present study is also an application for the visualisation of the structure of multi-layered neural networks whose solution is designed with respect to its potential utilization in the education of artificial intelligence. The first part of the study introduces the subject matter and formally describes both algorithms, followed by a short description of other variants of the algorithms and their analysis. The next part discusses the selection of the appropriate programming language for the implementation of the application, specifies the goals and describes the implementation works. The conclusion summarizes the test results of the speed and implementation comparison with the selected noncommercial-based software ENCOG.
|
|
|
Artificial neural networks and their application for 3D-data processing
Pihera, Josef ; Mrázová, Iveta (advisor) ; Holan, Tomáš (referee)
Neural networks represent a powerful means capable of processing various multi-media data. Two applications of artificial neural networks to 3D surface models are examined in this thesis - detection of significant features in 3D data and model classification. The theoretical review of existing self-organizing neural networks is presented and followed by description of feed-forward neural networks and convolutional neural networks (CNN). A novel modification of existing model - N-dimensional convolutional neural networks (ND- CNN) - is introduced. The proposed ND-CNN model is enhanced by an existing technique for enforced knowledge representation. The developed theoretical methods are assessed on supporting experiments with scanned 3D face models. The first experiment focuses on automatic detection of significant facial features while the second experiment performs classification of the models by their gender using the CNN and ND-CNN.
|
|
|
Handwritten character clustering
Novák, Jiří ; Štanclová, Jana (advisor) ; Mrázová, Iveta (referee)
This diploma thesis deals with comparision of di erent method for handwritten character recognition. A goal is analysis of this problem, comparision of several pattern classi cation methods and methods for feature extraction. In details there are discussed three algorithms for data clustering. These algorithms are following: k-means clustering, k-neighbors algorithm and algorithm of iterative optimalization. Another three algorithms are using neural networks. These are algorithm of competition learning, back propagation algorithm and Kohonen selforganising maps. These method are tested and compared according to percentage of correctly recognized characters. A summary of our own results is included.
|
|
|
Prototype for specialized business intelligence in merchandising
Jakab, Gergely ; Král, Jaroslav (advisor) ; Mrázová, Iveta (referee)
One of the purposes of merchandising is to place the correct product to the correct place with the correct price and at the right time. This thesis analyses the possibilities for automation of the process of product placement in racks of retail stores to help the merchandiser's work. The solution uses statistical data from sales to evaluate the products in order to choose the best sellers, then it suggests a suitable placement for the chosen products. Based on the results of the general analysis part a system is designed for automation of the process as an extension to an existing software. According to the created design a prototype is implemented.
|
|
|
Dynamic Kohonen maps and their strusture
Křižka, Radek ; Mrázová, Iveta (advisor) ; Sýkora, Ondřej (referee)
My diploma thesis deals with one of the most widely used model of artificial neural network named self-organizing Kohonen neural network. We can find there a detailed description of several thoroughly analyzed mutually compared models of Kohonen map. We will verify their functionality, robustness and generalisation rates on artificial input data. Their real applicability and properties are tested on real data of traffic accident frequency. We will focus on the detection of significant input data attributes. The possibilities of solving the interesting questions and aspects of road transport are examined by means of Kohonen maps. At the end of the work there is presented a summarized review of the results and there are mentioned possible options of modifications that could improve the properties of these models.
|
guest ::
RSS feed.