|
|
Symbolic Regression and Coevolution
Drahošová, Michaela ; Žaloudek, Luděk (referee) ; Sekanina, Lukáš (advisor)
Symbolic regression is the problem of identifying the mathematic description of a hidden system from experimental data. Symbolic regression is closely related to general machine learning. This work deals with symbolic regression and its solution based on the principle of genetic programming and coevolution. Genetic programming is the evolution based machine learning method, which automaticaly generates whole programs in the given programming language. Coevolution of fitness predictors is the optimalization method of the fitness modelling that reduces the fitness evaluation cost and frequency, while maintainig evolutionary progress. This work deals with concept and implementation of the solution of symbolic regression using coevolution of fitness predictors, and its comparison to a solution without coevolution. Experiments were performed using cartesian genetic programming.
|
|
|
Evolutionary Design of Image Classifier
Koči, Martin ; Bidlo, Michal (referee) ; Drahošová, Michaela (advisor)
This thesis deals with evolutionary design of image classifier with help of genetic programming, specifically with cartesian genetic programming. Thesis discribes teoretical basics of machine learing, evolutionary algorithms and genetic programming. Part of this thesis is described design of the program and its implementation. Futhermore, experiments are performed on two solved tasks for the classification of handwritten digits and the classification of cube drawings, which can be used to determine the rate of dementia in Parkinson's disease. The best designed solution for digits is with AUC of 0.95 and for cubes 0.86. Designed solutions are compared by other methods, namely convolutional neural networks (CNN) and the support vector machines (SVM). The resulting AUC for the classification of digits for both CNN and SVM is 0.99, for cubes CNN has a final AUC 0.81 and SVM 0.69. The cubes are then compared with existing solution, which resulted in AUC 0.70, so that the results of the experiments show an improvement in the method used in this thesis.
|
|
|
Coevolutionary Algorithms Statistical Analysis Tool
Urban, Daniel ; Zachariášová, Marcela (referee) ; Drahošová, Michaela (advisor)
This bachelor thesis contains a theoretical basis that introduces evolutionary algorithms, genetic programming, coevolutioanary algorithms and methods for statistical evaluation. Furthermore, this work deals with the design and implementation of tool with graphical user interface, which allows the analysis of coevolutioanary algorithm for various parameters and also its statistical evaluation. The functionality of the implemented tool has been tested on data obtained from an external program performing evolutionary design of image filters with the use of the coevolution of tness predictors. The resulting graphs and statistics allow easy comparison of the progress and results for each program run.
|
|
|
Competitive Coevolution in Cartesian Genetic Programming
Skřivánková, Barbora ; Petrlík, Jiří (referee) ; Drahošová, Michaela (advisor)
Symbolic regression is a function formula search approach dealing with isolated points of the function in plane or space. In this thesis, the symbolic regression is performed by Cartesian Genetic Programming and Competitive Coevolution. This task has already been resolved by Cartesian Genetic Programming using Coevolution of Fitness Predictors. This thesis is concerned with comparison of Coevolution of Fitness Predictors with simpler Competitive Coevolution approach in terms of approach effort. Symbolic regression has been tested on five functions with different complexity. It has been shown, that Competitive Coevolution accelerates the symbolic regression task on plainer functions in comparison with Coevolution of Fitness Predictors. However, Competitive Coevolution is not able to solve more complex functions in which Coevolution of Fitness Predictors succeeded.
|
|
|
Detection of Dynamic Network Applications
Burián, Pavel ; Drahošová, Michaela (referee) ; Kaštil, Jan (advisor)
This thesis deals with detection of dynamic network applications. It describes some of the existing protocols and methods of their identification from IP flow and packet contents. It constitues a design of a detection system based on the automatic creation of regular expressions and describes its implementation. It presents the created regular expressions for BitTorrent and eDonkey protocol. It compares their quality with the solution of L7-filter.
|
|
|
Coevolution of Image Filters and Fitness Predictors
Trefilík, Jakub ; Hrbáček, Radek (referee) ; Drahošová, Michaela (advisor)
This thesis deals with employing coevolutionary principles to the image filter design. Evolutionary algorithms are very advisable method for image filter design. Using coevolution, we can add the processes, which can accelerate the convergence by interactions of candidate filters population with population of fitness predictors. Fitness predictor is a small subset of the training set and it is used to approximate the fitness of the candidate solutions. In this thesis, indirect encoding is used for predictors evolution. This encoding represents a mathematical expression, which selects training vectors for candidate filters fitness prediction. This approach was experimentally evaluated in the task of image filters for various intensity of random impulse and salt and pepper noise design and the design of the edge detectors. It was shown, that this approach leads to adapting the number of target objective vectors for a particular task, which leads to computational complexity reduction.
|
|
|
Prediction of Secondary Structure of Proteins Using Cellular Automata
Brigant, Vladimír ; Drahošová, Michaela (referee) ; Bendl, Jaroslav (advisor)
This work describes a method of the secondary structure prediction of proteins based on cellular automaton (CA) model - CASSP. Optimal model and CA transition rule parameters are acquired by evolutionary algorithm. Prediction model uses only statistical characteristics of amino acids, so its prediction is fast. Achieved results was compared with results of other tools for this purpose. Prediction cooperation with a existing tool PSIPRED was also tested. It didn't succeed to beat this existing tool, but partial improvement was achieved in prediction of only alpha-helix secondary structure motif, what can be helful if we need the best prediction of alpha-helices. It was developed also a web interface of designed system.
|
|
|
Comparison of Genetic Programming Variants in the Symbolic Regression Task
Doležal, Petr ; Hurta, Martin (referee) ; Drahošová, Michaela (advisor)
This thesis deals with comparison of genetic programming variants it the task of symbolic regression. Time to converge and quality of evolved solutions are evaluated on nine chosen benchmarks. In particular, tree-based genetic programming, cartesian genetic programming and their modifications using coevolutionary algorithm are investigated. An own implementation of employed methods (without a specific library use) allows to share as much code as possible. Moreover, an analysis of implemented methods efficiency on real world data is provided. Experimental results show that all of the investigated approaches are capable of finding solutions using symbolic regression. Cartesian genetic programming enhanced with coevolution seems to be the most suitable of the investigated approaches in terms of evolved solution quality and time to converge.
|
|
|
Automatic Harmony Generation
Bobčík, Martin ; Drahošová, Michaela (referee) ; Vašíček, Zdeněk (advisor)
Goal of this master thesis is to study harmonization based on knowledge of given melody and to design a system which will meaningfully automate this activity. In the work there is covered basics of music theory needed for this topic and previous other approaches to this problematic. There is also covered machine learning, neural networks and recurrent neural networks. In the end, there is outlined design of the system, how to make it work and how to use it. Four experiments were executed with the system. Harmonization of the short melodies were unpleasant. Harmonization of longer melodies were overall more successful though. A possible cause might be relatively small used neural network of the system.
|
|
|
Coevolutionary Algorithm in FPGA
Hrbáček, Radek ; Vašíček, Zdeněk (referee) ; Drahošová, Michaela (advisor)
This thesis deals with the design of a hardware acceleration unit for digital image filter design using coevolutionary algorithms. The first part introduces reconfigurable logic device technology that the acceleration unit is based on. The theoretical part also briefly characterizes evolutionary and coevolutionary algorithms, their principles and applications. Traditional image filter designs are compared with the biologically inspired design methods. The hardware unit presented in this thesis exploits dual MicroBlaze system extended by custom peripherals to accelerate cartesian genetic programming. The coevolutionary image filter design is accelerated up to 58 times. The hardware platform functionality in the task of impulse noise filter design and edge detector design has been empirically analyzed.
|
guest ::
RSS feed.