|
|
Artificial intelligence for Texas Holdem poker game
Moravčík, Matej ; Petříčková, Zuzana (advisor) ; Sýkora, Ondřej (referee)
Recently there has been a great expansion of poker. This includes live games, as well as games on the internet. For beginners, it may be difficult to find opponents skilled enough and thus improve their gaming performance without deposit of their own funds. Using of artificial intelligence seems as good solution for the problem, but there are only few suitable programs available. This thesis describes the overall design and development of such an application, specially designed for tournament variant of Texas Hold'em poker. Most attention is devoted to the artificial intelligence. There are two main approaches discussed - approximate Nash equilibrium and the use of expert system. Emphasis is placed on the first option. The main contribution of this thesis is detailed description and comparison of three algorithms for calculating the approximation of Nash equilibrium. Two of them are original heuristics algorithms, that take advantage of specific structure of poker game. Algorithms have been implemented and their properties have been empirically evaluated. The final result is a full-featured application designed for end users. It simulates poker game and provides a powerful artificial intelligence with attractive graphical user interface.
|
|
|
Technical analysis of stock trends using artificial neural networks
John, Pavel ; Petříčková, Zuzana (advisor) ; Pilát, Martin (referee)
Although the discipline has not received the same level of acceptance in the past, the technical analysis has been part of financial practice for centuries. One of the big issues was the absence of widely respected fully rational background that is necessary for the modern science. The presence of geometrical shapes recognized by a human eye in historical data charts remained as one of the most important tools till the last decades. Nowadays, it is possible to find commercial trading software which employs neural networks. However, a freely accessible tool is difficult to obtain. The aim of this work was to investigate the usability of applications of neural networks on the technical analysis and to develop a software tool that would implement the knowledge acquired. An application was created and a new promising trading strategy proposed along with experimental data. The advantages of the program presented include the ease of extensibility and a high variability in trading strategies setting.
|
|
|
Gradient learning for networks of smoothly pulse neurons
Hošek, Lukáš ; Šíma, Jiří (advisor) ; Petříčková, Zuzana (referee)
Networks of spiking neurons present a biologically more plausible alternative to perceptron networks, having great potential for processing time series. However, as of now, no practically usable learning algorithm has been known. SpikeProp, based on a gradient descent method, and its modifications have a fundamental problem with dis-continuity of spike creation and deletion. A new nontrivial gradient learning algorithm for a model of smoothly spiking neurons is proposed as a possible way to solve this problem. The goal of this work is to implement and test this model and eventually propose further improvements.
|
|
|
Room Arrangement Generation
Dvořák, Ondřej ; Petříčková, Zuzana (referee) ; Jelínková, Eva (advisor)
The aim of this thesis is to create a program which generates layouts of furniture in a room on the basis of specific constraints. This program is called Spaceout and it is based on genetic algorithms. The thesis also gives a basic overview of existing programs for creating projects of interiors and exteriors. The thesis contains a programmer and user documentation of Spaceout.
|
|
|
Improving and extending the multiple sequence alignment suite PRALINE
Hudeček, Jan ; Petříčková, Zuzana (referee) ; Mráz, František (advisor)
The aim of this work is to study potential improvements in the core routines of multiple sequence alignment suite PRALINE. A general overview of multiple sequence alignment methods used with emphasis on representation of the alignment core is given. A new option for aligning sequence profiles was implemented and its usefulness assessed. This option allows a user to input a profile which is used in an advanced phase of the progressive protocol as if it was a result of the previous steps. Two new protocols using profile Hidden Markov models (HMM) and their alignment were implemented and tested. The HMMGUIDE protocol creates for each sequence a preprofile consisting of segments of other sequences with high local similarity. HMM is generated from each preprofile by HMMER, and alignment of every pair is scored by PRC. The protocol then progressively aligns the sequence whose HMMs achieved the best score. The PRCALIGN protocol works similarly but aligns the sequences according to the best alignment of the HMMs. While not all test alignments were finished successfully for both protocols, the results constitute a statistically significant improvement over the original PRALINE protocol.
|
|
|
Modely výpočetní inteligence pro hydrologické predikce
Paščenko, Petr ; Petříčková, Zuzana (referee) ; Neruda, Roman (advisor)
The thesis deals with the application of computational artificial intelligence models on hydrological predictions. The short term rainfall-runoff prediction problem is studied on the real data of physical time seriesmeasured in the watershed of river Plučnice. A brief statistical study including correlation and regression analyses is performed. The high level of variance and noise is concluded. The evolution of the proper input filter providing an input set for the neural network is performed. In the main part of the thesis several neural network models based on multilayer perceptron, RBF units, and neuroevoution are constructed together with two neural ensembles inspired by the bagging method. The models are tested on the three subsequent years summer data. The greater generalization ability of multilayer perceptron architectures is concluded. The resulting multilayer perceptron models are able to reduce the mean squared error of the prediction by 15% compared to the prediction by the previous value.
|
|
|
Knowledge Extraction from Data
Kozák, Vladislav ; Petříčková, Zuzana (referee) ; Mrázová, Iveta (advisor)
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.
|
|
| |
|
|
Development of robotic soccer players by methods of evolutionary programming
Dener, Libor ; Petříčková, Zuzana (referee) ; Mráz, František (advisor)
In the present work we study and implmement means to evolve players of robotic soccer. We implement our own simplied model of soccer play and libraries for genetic algorithms and artical neural networks. We focus on players controlled by artical neural networks in the simplied model. We use incremental learning approach in which we rst train the players on simpler subtasks and then we use these players on more complex problems. The neural networks are evolved by genetic algorithms. We demonstrate dierent variants of genetic algorithms and we discuss achieved results.
|
|
|
Multilayer hierarchical models
Béger, Michal ; Petříčková, Zuzana (referee) ; Štanclová, Jana (advisor)
This diploma thesis deals with hierarchical associative memories (HAM), which have been experimentally analysed only in the case of two layer hierarchy so far. The aim of this thesis is to study existing hierarchical models and evaluate experimentally their performance for more than two layer hierarchy. We show, that existing hierarchical model HAM is not suitable for three or more layer hierarchy. For that reason, we propose a new version of hierarchical model (called HAM-N), which enables utilization of any number of layers. The new model HAM-N uses the structure of the HAM model. However, due to modied learning and recall process, the HAM-N model eliminates the above-mentioned drawbacks of the HAM model. Finally, the HAM-N model is experimentally studied with respect to processing of large amounts of correlated patterns. Thesis also includes analysis of experiment results.
|
guest ::
RSS feed.