|
|
Clustering hits and predictions in data from TimePix3 detectors
Čelko, Tomáš ; Mráz, František (advisor) ; Holan, Tomáš (referee)
Hybrid pixel detectors like Timepix3 and Timepix4 detect individual pixels hit by particles. For further analysis, individual hits from such sensors need to be grouped into spatially and temporally coinciding groups called clusters. While state-of-the-art Timepix3 detectors generate up to 80 Mio hits per second, the next generation, Timepix4, will provide data rates of up to 640 Mio hits, which is far beyond the current capabilities of the real-time clustering algorithms, processing at roughly 3 MHits/s. We explore the options for accelerating the clustering process, focusing on its real-time application. We developed a tool that utilizes multicore CPUs to speed up the clustering. Despite the interdependence of different data subsets, we achieve a speed-up scaling with the number of used cores. Further, we exploited options to reduce the computational demands of the clustering by determining radiation field parameters from raw (unclustered) data features and automatically initiating further clustering if these data show signs of interesting events. This further accelerates the clustering while also reducing storage space requirements. The proposed methods were validated and benchmarked using real-world and simulated datasets.
|
|
|
Evolution of robots in a simulated physical environment
Bečvář, Marek ; Mráz, František (advisor) ; Vodrážka, Jindřich (referee)
This work introduces a system for designing and evaluating experiments with evo- lutionary algorithms in 3D-simulated physical environments of the MuJoCo library. Ex- periments allow to develop the control and morphology of robots while using arbitrary user-defined fitness functions. The implementation was designed to be as accessible, un- derstandable, and extendable as possible. The system offers a simple graphical user in- terface allowing a detailed configuration of experiments and a text-based user interface which is convenient for running large amounts of experiments for statistical analysis. The work implements several robots of different complexity, examples of various evolutionary algorithms, and a selection of well-known genetic operators. During experiment design, the architecture of this system allows the combining of implemented operators and tools arbitrarily. This work and the user documentation give simple instructions on how to alter and extend the implementation. 1
|
|
|
Learning picture languages using picture-to-string transformations
Rychtera, Ivan ; Mráz, František (advisor) ; Pešková, Klára (referee)
We define a new model for recognizing picture languages using picture-to-string trans- formations - transcriptor-evaluator machine for picture languages (TEMPL). The model consists of a transcriptor that rewrites an input picture into a string and an evaluator that accepts or rejects the obtained string. We use TEMPL as a representation of pic- ture languages that can be learned from positive and negative examples. The attached implementation of TEMPL and several algorithms for learning regular languages were used for benchmarking the accuracy of training TEMPLs on selected picture languages. 1
|
|
| |
|
|
Testing Learning of Restarting Automata using Genetic Algorithm
Kovářová, Lenka ; Mráz, František (advisor) ; Černo, Peter (referee)
Title: Testing the Learning of Restarting Automata using Genetic Algorithm Author: Bc. Lenka Kovářová Department: Department of Software and Computer Science Education Supervisor: RNDr. František Mráz, CSc. Abstract: Restarting automaton is a theoretical model of device recognizing a formal language. The construction of various versions of restarting automata can be a hard work. Many different methods of learning such automata have been developed till now. Among them are also methods for learning target restarting automaton from a finite set of positive and negative samples using genetic algorithms. In this work, we propose a method for improving learning of restarting automata by means of evolutionary algorithms. The improvement consists in inserting new rules of a special form enabling adaption of the learning algorithm to the particular language. Furthermore, there is proposed a system for testing of learning algorithms for restarting automata supporting especially learning by evolutionary algorithms. A part of the work is a program for learning restarting automata using the proposed new method with a subsequent testing of discovered automata and its evaluation in a graphic form mainly. Keywords: machine learning, grammatical inference, restarting automata, genetic algorithms
|
|
|
Busy Beaver Problem
Kropitz, Pavel ; Holan, Tomáš (advisor) ; Mráz, František (referee)
The work's purpose was to develop and implement optimalization methods that could be used for solving Busy Beaver problem with order of 5+. The result of the work is the theoretical part and its implementation in form of two programs - simulator of Turing Machines which shows the computation of the machine in detail along with application of theory, and program searching the space of Turing Machines. The latter was ran for machines with four to six states. The quality of the methods was proven by small number of machines that the program could not detect and by nding a new record machine - candidate for 6-state Busy Beaver.
|
|
|
Generation of images and animations using L-systems
Jaška, Milan ; Mráz, František (advisor) ; Holan, Tomáš (referee)
Main goal of this thesis is to design system for editing and visualizing planar and cubical scenes consisting of elementary objects and objects described by Lindenmayers systems. Planar scenes visualization will be straight and cubical scenes visualization will be done through VRML language. VRML data will be visualized using external program. System should be able to capture chosen stages of Lindenmayer objects development as pictures so that it is possible to animate the development of object.
|
|
|
Visualization and testing of team profiles
Höschl, Cyril ; Mráz, František (advisor) ; Holan, Tomáš (referee)
The aim of this thesis is to demonstrate methods for visualization and testing team profiles. An individual's profile is a vector of scores representing personality of a team member. Team profile is a collection of individual profiles. The first step for visualization of team profile is to find similarities among individual profiles. We present new approach for measuring similarity between two profiles as a significance of difference expressed by probability. We use this measure both for testing mean difference between two subgroups of profiles and also for building similarity matrix. We represent team profiles as a 3D ladscape map called Sociomap. To create a sociomap we demonstrate own algorithm for projecting similarity matrix into a 2D plane. We compare this algorithm with well known Multidimensional scaling and Principal Component Analysis methods solving the same task. Comparing with MDS and PCA, our algorithm provides more accurate result with comparable demand of time. We also employ some well known Computer Graphics methods to improve visual 3D representation of the sociomap and we also show some new improvements to render sociomap more realistic.
|
|
|
Evolutionary development of robotic organisms
Leibl, Marek ; Mráz, František (advisor) ; Holan, Tomáš (referee)
This work introduces a system for an evolutionary design of virtual organisms capable of effective movement in a simulated environment. The morphology and the control system are simultaneously developed by an evolutionary algorithms. The system also allows to design organisms in an editor and evolution of the control system with an immutable morphology. The quality evaluation and viewing of evolved organisms is done in a simulated 3D physical environment. The work put stress on the optimization of time and computing complexity of the evolutionary process. This optimization is achieved by using symmetry of organisms and their movement with HyperNEAT-generative encoding of synaptic values. Further optimization is achieved by limiting the variety of mutual module connections and focusing on the harmonic movement of organisms.
|
|
| |
guest ::
