|
|
Using genetic programming in robot evolution
Babor, Petr ; Mráz, František (advisor) ; Neruda, Roman (referee)
Artificial neural networks learned by evolutionary algorithms are commonly used to control the robots. Neural networks can be encoded either directly as a list of weights or indirectly as a weight generator. Unlike direct coding indirect encoding allows to encode a large network using a short genetic code. HyperNEAT is a neuroevolutionary algorithm, which encodes the neural network indirectly, through another (producing) network, which computes synaptic weights. A different algorithm called HyperGP is an alternative to HyperNEAT. In HyperGP, the producing network is replaced by an arithmetic expression, which is being evolved using a genetic programming (GP). We have designed enhancements for HyperGP, using techniques that are either known in a different context of GP or completely new. Algorithm and enhancements have been implemented and experimentally tested on a task of controlling virtual walking robot. The results were compared with HyperNEAT and with the original HyperGP. We have shown that most of the proposed enhancements are effective and, on the given task, HyperGP is better than HyperNEAT. GP thus can successfully replace NEAT in hyper-encoding scheme and improve its efficiency. Powered by TCPDF (www.tcpdf.org)
|
|
|
Alhambra
Klůj, Jan ; Holan, Tomáš (advisor) ; Mráz, František (referee)
Title: Alhambra Author: Jan Klůj Department: Department of Software and Computer Science Education Supervisor: RNDr. Tomáš Holan, Ph.D., Department of Software and Computer Science Education Abstract: The bachelor thesis deals with the implementation of the board game Alhambra. Besides of the implementation of the game rules, program includes also a graphical user interface. The game can be played by two to six players who take turns at one computer. Further we deal with an artificial intelligence, against which we can play. Decision logic of the artificial intelligence is made by using the evolution algorithm and machine learning. Keywords: Alhambra, game, artificial intelligence, evolution algorithm
|
|
|
Cellular Automata
Amemori, Josef ; Barták, Roman (advisor) ; Mráz, František (referee)
A Cellular automaton is a simple mathematical model that can exhibit a complex behavior. It was introduced by Von Neumann who was trying to find a mathematical description for a self-reproduction. Among well known works about self-reproduction belongs works from Hiroky Sayama. He introduced a model, that was capable of a simple evolution. This work expands Sayama`s evolving model by simple interactions inspired by competitions of specieces. The aim of the work is observe better dynamics than in Sayama`s model.
|
|
|
Modular and ontogenetic evolution of virtual organisms
Leibl, Marek ; Mráz, František (advisor) ; Šmíd, Jakub (referee)
Increase of computational power and development of new methods in artificial intelligence allow these days many real-world problems to be solved automatically by a~computer program without human interaction. This includes automatized design of walking robots in a~physical virtual environment that can eventually result in construction of real robots. This work compares two different approaches to evolve virtual robotic organisms: artificial ontogeny, where the organism first grows using an~artificial ontogenetic process, and more direct methods. Furthermore, it proposes a~novel approach to evolve virtual robotic organisms: Hypercube-based artificial ontogeny that is combination of artificial ontogeny and Hypercube-based neuroevolution of augmenting topologies (HyperNEAT). Powered by TCPDF (www.tcpdf.org)
|
|
|
Návrh efektivní generické molekulární reprezentace
Škoda, Petr ; Hoksza, David (advisor) ; Mráz, František (referee)
The screening of chemical libraries is an important step in the drug discovery process. The existing chemical libraries contain up to millions of compounds. As the screening at such scale is expensive, the virtual screening is often utilized. There exist several variants of virtual screening and ligand- based virtual screening is one of them. It utilizes the similarity of screened chemical compounds to known compounds. Besides the employed similarity measure, another aspect greatly influencing the performance of ligand-based virtual screening is the chosen chemical compound representation. In this thesis, we introduce a fragment-based representation of chemical compounds. Our representation utilizes fragments to represent a compound. Each fragment is represented by its physicochemical descriptors. The representation is highly parameterizable, especially in the area of physicochemical descriptors selection and application. In order to test the performance of our method, we utilized an existing framework for virtual screening benchmarking. The results show that our method is comparable to the best existing approaches and on some datasets it outperforms them.
|
|
|
Learning picture languages using restarting automata
Krtek, Lukáš ; Mráz, František (advisor) ; Průša, Daniel (referee)
There are many existing models of automata working on two-dimensional inputs (pictures), though very little work has been done on the subject of learning of these automata. In this thesis, we introduce a new model called two-dimensional limited context restarting automaton. Our model works similarly as the two-dimensional restarting tiling automaton, yet we show that it is equally powerful as the two-dimensional sgraffito automaton. We propose an algorithm for learning of such automata from positive and negative samples of pictures. The algorithm is implemented and subsequently tested with several basic picture languages. Powered by TCPDF (www.tcpdf.org)
|
|
| |
|
|
NetHack Bot Framework
Krajíček, Jan ; Gemrot, Jakub (advisor) ; Mráz, František (referee)
Previous attempts at implementing bots for the classic roguelike game NetHack have been hindered by many problems related to its complexity and console-based interface. The framework implemented as part of this work solves the problem of interfacing with the game and provides a programmer-friendly API for the Java and Clojure programming languages. It enables programming sophisticated bots using the provided model of the game world, a library of possible actions and utilities for various aspects of the game. The framework uses elements of functional and logic programming and doesn't require modifications of the game. Also described is an implementation of the first NetHack bot capable of winning the game. Powered by TCPDF (www.tcpdf.org)
|
|
|
Meta-learning methods for analyzing Go playing trends
Moudřík, Josef ; Neruda, Roman (advisor) ; Mráz, František (referee)
This thesis extends the methodology for extracting evaluations of players from samples of Go game records originally presented in (Baudiš - Moudřík, 2012). Firstly, this work adds more features and lays out a methodology for their comparison. Secondly, we develop a robust machine-learning framework, which is able to capture dependencies between the evaluations and general target variable using ensemble meta-learning with a genetic algorithm. We apply this framework to two domains, estimation of strength and styles. The results show that the inference of the target variables in both cases is viable and reasonably precise. Finally, we present a web application, which realizes the methodology, while presenting a prototype teaching aid for the Go players and gathering more data. Powered by TCPDF (www.tcpdf.org)
|
|
|
Protein Structure Similarity Using Genetic Programming
Šiagi, Miroslav ; Hoksza, David (advisor) ; Mráz, František (referee)
The thesis deals with the protein structure similarity problem which is an important aspect of bioinformatics. Due to exponential growth of protein structures in databases, the development of more effective methods is required. Principles of evolutionary computation offer a way to solve the similarity problem. We focus on one of the evolutionary paradigms - genetic programming. The main advantage of genetic programming is a tree representation. We propose new method called ProSSiGen using genetic programming. ProSSiGen is evaluated by automatic protein classification. Obtained results signify that the efficiency of our method is insufficient. Regardless of the inefficiency, there are many reasons to continue to research. One of the reasons is the capability of genetic programming.
|
guest ::
