|
|
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.
|
|
|
Coevolutionary Algorithms and Classification
Hurta, Martin ; Sekanina, Lukáš (referee) ; Drahošová, Michaela (advisor)
The aim of this work is to automatically design a program that is able to detect dyskinetic movement features in the measured patient's movement data. The program will be developed using Cartesian genetic programming equipped with coevolution of fitness predictors. This type of coevolution allows to speed up a design performed by Cartesian genetic programming by evaluating a quality of candidate solutions using only a part of training data. Evolved classifier achieves a performance (in terms of AUC) that is comparable with the existing solution while achieving threefold acceleration of the learning process compared to the variant without the fitness predictors, in average. Experiments with crossover methods for fitness predictors haven't shown a significant difference between investigated methods. However, interesting results were obtained while investigating integer data types that are more suitable for implementation in hardware. Using an unsigned eight-bit data type (uint8_t) we've achieved not only comparable classification performance (for significant dyskinesia AUC = 0.93 the same as for the existing solutions), with improved AUC for walking patient's data (AUC = 0.80, while existing solutions AUC = 0.73), but also nine times speedup of the design process compared to the approach without fitness predictors employing the float data type, in average.
|
|
|
Evolutionary Design of Boolean Functions for Cryptography
Dvořák, Jan ; Vašíček, Zdeněk (referee) ; Husa, Jakub (advisor)
The goal of this bachelor's thesis is to compare various selection methods used in cartesian genetic programming applied to a problem of various types of cryptographically significant boolean functions. I focused on these selection methods: evolutionary strategies (1+lambda) and (1,lambda), tournament selection and roulette selection. The chosen problem was solved by an implementation of CGP with the above-mentioned selection methods and by a statistical evaluation of data acquired from conducted experiments. Evaluation of mentioned data has shown that the best results in case of bent functions were achieved while using (1+lambda) evolutionary strategy. The roulette selection performed the best in case of balanced functions with high nonlinearity.
|
|
|
Design of S-Boxes Using Genetic Algorithms
Hovorka, Bedřich ; Zadina, Martin (referee) ; Hanáček, Petr (advisor)
This work deals with part of the encryption algorithm, called S-box and its development. For its development is used evolutionary computing, such as classical genetic algorithm, Estimation of Distribution Algorithm, Cartesian genetic programming and multi-criteria VEGA and SPEA algorithms. This thesis aims to test the properties of substitution boxes to its evolutionary development. Firstly, the work deals with cryptography and issues of s-boxes. There are explained basic concepts and describes the selected criteria of safety. Next chapter explains evolutionary algorithms and multi-criteria optimization. This knowledge is used to design and program implementation, which are described below. Finally discusses the application of the criteria studied. Discussed here is searching S-boxes in both single-criteria, and especially in multi-criteria genetic search.
|
|
|
Dynamic Approximation of Digital Circuits
Jásenský, Michal ; Hrbáček, Radek (referee) ; Sekanina, Lukáš (advisor)
This bachelor's thesis deals with design of a method based on cartesian genetic programming, which allows the evolutionary design of circuits capable of dynamic reconfiguration. The goal of reconfiguration is to dynamically change the number of used components and thereby to change the accuracy of calculation. In this thesis, implementation of the proposed method is described. The method is experimentally verified and demonstrated on several selected circuits.
|
|
|
Coevolution of Image Filters and Noise Detectors
Komjáthy, Gergely ; Zachariášová, Marcela (referee) ; Drahošová, Michaela (advisor)
This thesis deals with image filter design using coevolutionary algorithms. It contains a description of evolutionary algorithms, focusing on genetic programming, cartesian genetic programming and coevolution, the reader can learn about image filters too. The next chapters contain the design of image filters and noise detectors using cooperative coevolution, and the implementation and testing of the proposed filter. In the last chapter the proposed filter is compared to other filters created using evolutionary algorithms but without coevolution.
|
|
|
Geometric Semantic Genetic Programming
Končal, Ondřej ; Bidlo, Michal (referee) ; Sekanina, Lukáš (advisor)
This thesis examines a conversion of a solution produced by geometric semantic genetic programming (GSGP) to an instantion of cartesian genetic programming (CGP). GSGP has proven its quality to create complex mathematical models; however, the size of these models can get problematically large. CGP, on the other hand, is able to reduce the size of given models. This thesis combinated these methods to create a subtree CGP (SCGP). The SCGP uses an output of GSGP as an input and the evolution is performed using the CGP. Experiments performed on four pharmacokinetic tasks have shown that the SCGP is able to reduce the solution size in every case. Overfitting was detected in one out of four test problems.
|
|
|
Genetic Improvement of Cartesian Genetic Programming Software
Husa, Jakub ; Jaroš, Jiří (referee) ; Sekanina, Lukáš (advisor)
Genetic programming is a nature-inspired method of programming that allows an automated creation and adaptation of programs. For nearly two decades, this method has been able to provide human-comparable results across many fields. This work gives an introduction to the problems of evolutionary algorithms, genetic programming and the way they can be used to improve already existing software. This work then proposes a program able to use these methods to improve an implementation of cartesian genetic programming (CGP). This program is then tested on a CGP implementation created specifically for this project, and its functionality is then verified on other already existing implementations of CGP.
|
|
|
Movement Abnormalities Classification using Genetic Programming
Chudárek, Aleš ; Mrázek, Vojtěch (referee) ; Drahošová, Michaela (advisor)
When suppressing the symptoms of Parkinson's disease, the correct dosage of drugs is critical for the patient. Improper dosing can either cause insufficient suppression of symptoms or, conversely, side effects, such as dyskinesia, occur with high doses. Dyskinesia is manifested by involuntary muscle movement. This work deals with the automated classification of dyskinesia from motion data recorded using a triaxial accelerometer located on the patient's body. In this work, the classifier of dyskinesia is automatically designed using Cartesian genetic programming. The designed classifier achieves very good quality of classification of severe dyskinesia (AUC = 0,94), which is a comparable result to the techniques presented in scientific literature.
|
|
|
Acceleration of Symbolic Regression Using Cartesian Genetic Programming
Hodaň, David ; Mrázek, Vojtěch (referee) ; Vašíček, Zdeněk (advisor)
This thesis is focused on finding procedures that would accelerate symbolic regressions in Cartesian Genetic Programming. It describes Cartesian Genetic Programming and its use in the task of symbolic regression. It deals with the SIMD architecture and the SSE and AVX instruction set. Several optimizations that lead to a significant acceleration of evolution in Cartesian Genetic Programming are presented. A method of a bit-level parallel simulation that uses AVX2 vectors allows to process 256 input combinations of a logic circuit in paralell. Similarly it is possible to use a byte-level parallel simulation and work with 32 bytes when evolving an image filter. A new method of batch mutation can accelerate the evolution of combinational logic circuits thousand times depending on the problem size. For example, using a combination of these and other methods the evolution of 5 x 5b multipliers took 5.8 seconds on average on an Intel Core i5-4590 processor.
|
guest ::
