Preferencje help
Widoczny [Schowaj] Abstrakt
Liczba wyników

Znaleziono wyników: 15

Liczba wyników na stronie
first rewind previous Strona / 1 next fast forward last
Wyniki wyszukiwania
Wyszukiwano:
w słowach kluczowych:  evolutionary computing
help Sortuj według:

help Ogranicz wyniki do:
first rewind previous Strona / 1 next fast forward last
EN
The paper presents the application of a multi-objective P-Estra evolutionary algorithm to the improvement of indoor deployment of radio nodes/beacons serving both network access and positioning purposes. In the paper we introduce modifications to the objective function components formulation, aimed at simultaneous improvement of service coverage and indoor positioning accuracy. Simulation results illustrating the performance of the proposed method are shown an discussed.
PL
W artykule zaprezentowano zastosowanie algorytmu genetycznego P-EStra z dwukryterialną funkcją celu, do poprawy rozmieszczenia punktów dostępowych/radiolatarni służących zarówno do zapewnienia dostępu do sieci, jak i do celów wyznaczania lokalizacji terminali. W artykule zaproponowano modyfikacje składowych funkcji celu, służące jednoczesnej poprawie zarówno zasięgu sieci, jak i dokładności wyznaczania położenia. Przedstawiono i poddano dyskusji wyniki symulacji komputerowych ilustrujących działanie zaproponowanej metody.
2
Content available Fine tuning of agent-based evolutionary computing
EN
Evolutionary Multi-agent System introduced by late Krzysztof Cetnarowicz and developed further at the AGH University of Science and Technology became a reliable optimization system, both proven experimentally and theoretically. This paper follows a work of Byrski further testing and analyzing the efficacy of this metaheuristic based on popular, high-dimensional benchmark functions. The contents of this paper will be useful for anybody willing to apply this computing algorithm to continuous and not only optimization.
EN
The research applications of fuzzy logic have always been multidisciplinary in nature due to its ability in handling vagueness and imprecision. This paper presents an analytical study in the role of fuzzy logic in the area of metaheuristics using Web of Science (WoS) as the data source. In this case, 178 research papers are extracted from it in the time span of 1989-2016. This paper analyzes various aspects of a research publication in a scientometric manner. The top cited research papers, country wise contribution, topmost organizations, top research areas, top source titles, control terms and WoS categories are analyzed. Also, the top 3 fuzzy evolutionary algorithms are extracted and their top research papers are mentioned along with their topmost research domain. Since neuro fuzzy logic poses feasible options for solving numerous research problems, hence a section is also included by the authors to present an analytical study regarding research in it. Overall, this study helps in evaluating the recent research patterns in the field of fuzzy metaheuristics along with envisioning the future trends for the same. While on one hand this helps in providing a new path to the researchers who are beginners in this field as they can start exploring it through the analysis mentioned here, on the other hand it provides an insight to professional researchers too who can dig a little deeper in this field using knowledge from this study.
EN
Nowadays, more and more frequently, Wireless Local Area Networks are used not only for Internet access but also for indoor location services. The paper presents the utilization of an evolutionary computing implementation, called EStra, to the improvement of deployment of Wireless Local Area Network access nodes/radio beacons. In the paper, a new objective function is proposed for simultaneous improvement of coverage and indoor positioning accuracy. The function comprises a GDOP (HDOP) – Geometric (Horizontal) Dilution of Precision component, a concept originally used in satellite navigation systems as positional measurement precision. A new, novel, feature of the algorithm is a variable number of nodes so that their numbercan be minimized. Simulation results illustrating performance of the proposed method are shown.
PL
Współcześnie, coraz częściej bezprzewodowe sieci lokalne (WLAN) są wykorzystywane nie tylko dla zapewnienia dostępu do Internetu, ale także do realizacji usług lokalizacyjnych we wnętrzach budynków. W artykule zaprezentowano zastosowanie implementacji algorytmu genetycznego, nazwanej EStra, do optymalizacji rozmieszczenia punktów dostępowych/radiolatarni WLAN. Zaproponowano funkcję celu, która umożliwia jednoczesną optymalizację zasięgu sieci WLAN i dokładności lokalizacji, między innymi dzięki włączeniu w nią miary GDOP (HDOP) – Geometric (Horizontal) Dilution of Precision, oryginalnie wykorzystywanej w systemach nawigacji satelitarnej. Oryginalnym i nowym rozwiązaniem jest dodatkowo wersja algorytmu EStra z dynamicznie zmienną liczbą parametrów optymalizacji.
PL
W artykule przedstawiono wykorzystanie algorytmu ewolucyjnego (Estra) do identyfikacji parametrów uproszczonego modelu ludzkiego ciała (fantomu). Uproszczony model ciała może być wykorzystywany do symulacji odstrojenia impedancyjnego anteny znajdującej się w pobliżu ciała. W artykule przedstawiono sposób określenia parametrów uproszczonego modelu za pomocą automatycznej procedury opartej na algorytmie ewolucyjnym i metodzie różnic skończonych w dziedzinie czasu (FDTD). Po określeniu wartości parametrów, uproszczony model został porównany do heterogenicznego modelu ludzkiego ciała. Modele porównano w oparciu o analizę dopasowania impedancyjnego anteny dipolowej znajdującej się na obu modelach.
EN
The paper presents the exploitation of a lowestorder algorithm of evolutionary computing (EStra) for identifying the parameters of a simplified human body model (phantom). A simplified model is well suited in view of the computationally-expensive field simulation of wearable antennas located in a close proximity to the human body. In the paper, an automated procedure based on evolutionary computing and Finite Difference Time Domain (FDTD) computational electrodynamics method is proposed to identify the parameters of the simplified model. Subsequently, after identifying the parameter values, the simplified model is compared to a heterogeneous anthropomorphic human-body model. The comparison is based on the analysis of impedance matching of the same dipole antenna located on both the anthropomorphic and simplified phantoms.
EN
The paper presents the continuation of the author’s research on evolutionary approach to ship trajectory planning. While the general problem of the evolutionary trajectory planning has already been solved, no one has yet touched one of its specific aspects: evolutionary trajectory planning within Traffic Separation Schemes. Traffic Separation Scheme (TSS) is a traffic-management route-system complying with rules of the International Maritime Organization. In brief, the ships navigating within a TSS all sail in the direction assigned to a particular traffic lane or they cross at a course angle as close to 90 degrees as possible. This and other TSS rules largely affect the evolutionary method. The paper presents a proposal of the extended evolutionary method, with a focus on changes that have to be made to obey TSS rules, especially the changes in the phases of evaluation and specialised operators of the evolutionary cycle.
EN
This paper concerns efficient parameters tuning (meta-optimization) of a state-of-the-art metaheuristic, Quantum-Inspired Genetic Algorithm (QIGA), in a GPU-based massively parallel computing environment (NVidia CUDATMtechnology). A novel approach to parallel implementation of the algorithm has been presented. In a block of threads, each thread transforms a separate quantum individual or different quantum gene; In each block, a separate experiment with different population is conducted. The computations have been distributed to eight GPU devices, and over 400× speedup has been gained in comparison to Intel Core i7 2.93GHz CPU. This approach allows efficient meta-optimization of the algorithm parameters. Two criteria for the meta-optimization of the rotation angles in quantum genes state space have been considered. Performance comparison has been performed on combinatorial optimization (knapsack problem), and it has been presented that the tuned algorithm is superior to Simple Genetic Algorithm and to original QIGA algorithm.
EN
In this paper, implementation of Quantum-Inspired Genetic Algorithm(QIGA) in massively parallel environment (Graphics Processing Units) has been presented. Contrary to many recent papers concerning parallel implementation of evolutionary algorithms, in this paper a novel approach has been taken. QIGA algorithm has been implemented entirely as a computational kernel. Parallelization of the algorithm has been performed on two levels: In a block of threads, each thread transforms a separate individual or different gene; In each block, separate populations with same or different parameters are evolved. Finally, the computations have been distributed to eight GPU devices, and over 400x speedup has been gained in comparison to sequential implementation of the algorithm in ANSI C on one Intel Core i7 2.93 GHz CPU core. Correctness of the results has been verified in statistical analysis. The presented approach can be applied to experimentation with a broad class of metaheuristics.
PL
W artykule zostały przedstawione szczegóły implementacji kwantowo inspirowanego algorytmu genetycznego (QIGA) w środowisku obliczeń masowo równoległych na procesorach kart graficznych. W odróżnieniu od wielu dotychczasowych opracowań, prezentujących implementacje algorytmów ewolucyjnych w środowiskach obliczeń równoległych, w niniejszym artykule zostało zaproponowane nowatorskie podejście do implementacji algorytmu ewolucyjnego. Zrównoleglenie algorytmu zostało wykonane na dwóch poziomach: poszczególne osobniki w populacji lub poszczególne geny są przetwarzane przez osobne wątki w blokach, a w poszczególnych blokach przeprowadzany jest proces ewolucji populacji o tych samych lub różnych parametrach. Obliczenia zostały rozdzielone na osiem jednostek GPU, co pozwoliło na uzyskanie ponad 400-krotnego przyśpieszenia algorytmu w stosunku do sekwencyjnej implementacji w języku ANSI C na pojedynczym rdzeniu procesora Intel Core i7 2,93 GHz. Poprawność implementacji została zweryfikowana poprzez analizę statystyczną otrzymanych wyników. Zaproponowane podejście pozwala przyśpieszyć badanie dowolnych metaheurystyk przeszukiwania.
EN
This paper presents a comparison of selected algorithms for simultaneous localization and mapping (SLAM) problem in mobile robotics. Results of four general metaheuristics, Simple Genetic Algorithm, Particle Swarm Optimization, Quantum-Inspired Genetic Algorithms and Genetic Algorithm with Quantum Probability Representation, have been compared to results of classical, analytic method in this field, Iterative Closes Points algorithm. In the experiments the same objective function, drawn from Iterative Closest Points algorithm, has been used. Two situations have been considered: local and global localization problems of mobile robot. Both problems are import and often critical for successful navigation of robot in environment.
PL
W artykule zostało przedstawione porównanie wybranych algorytmów w zadaniu lokalizacji w przestrzeni robota mobilnego. Poddane analizie zostały wyniki uzyskane przez cztery ogólne metaheurystyki przeszukiwania: klasyczny algorytm genetyczny, metoda roju cząstek oraz dwa kwantowo inspirowane algorytmy genetyczne. Wyniki zostały porównane z klasyczną, analityczną metodą Iterative Closest Points, wykorzystywaną często do rozwiązywania rozważanego w artykule problemu. We wszystkich eksperymentach została wykorzystana taka sama funkcja celu, utworzona przy wykorzystaniu algorytmu Iterative Closest Points. Rozważono dwa warianty zagadnienia lokalizacji: problem lokalizacji lokalnej oraz globalnej. Obydwa zagadnienia mają krytyczne znaczenie w prawidłowym funkcjonowaniu autonomicznego robota mobilnego.
10
Content available Memetic algorithm for assembly sequence planning
EN
The paper presents the application of a memetic algorithm to searching for the optimal sequence of the assembly of parts. Such approach is based on the use of an algorithm connecting two methods of global and local search in order to increase the effectiveness of the conducted optimisation process. Based on a proper representation of assembly sequences and a set of geometrical, topological and technological constraints, connected with the attributes of a product, it is possible to create an evolutionary model. Through proper control of the evolution process in a model, based on the appropriate selection of parameters, it is possible to achieve good results in a short period of time. Although the evolutionary algorithm does not guarantee the obtaining of optimal solutions, it has been proven, based on sample simulations, that such solutions are obtained in a repeated manner. The application of the presented evolutionary approach enables creating fast assembly sequence planning tools, indispensable in tactical planning and operational control of manufacturing processes.
11
Content available remote Non-dominated Rank based Sorting Genetic Algorithms
EN
In this paper a new concept of ranking among the solutions of the same front, along with elite preservation mechanism and ensuring diversity through the nearest neighbor method is proposed for multi-objective genetic algorithms. This algorithm is applied on a set of benchmark multi-objective test problems and the results are compared with that of NSGA-II (a similar algorithm). The proposed algorithm is seen to over perform the existing algorithm. More specifically, the new approach has been used to solve the deceptive multi-objective optimization problems in a better way.
12
Content available remote Immune and evolutionary shape optimization in forgin
EN
The paper deals with applications of methods of artificial intelligence: artificial immune systems and evolutionary algorithms in optimization of a forging process. The shape optimization of the anvils in a two-stage forging process is considered as a numerical example. The paper contains description of the evolutionary algorithm, the artificial immune system and parallel versions of bioinspired algorithms in grid environment.
PL
W artykule przedstawiono zastosowanie dwóch biologicznie inspirowanych metod obliczeniowych - algorytmów ewolucyjnych i sztucznych systemów immunologicznych w optymalizacji procesu kucia. Dwu-etapowy proces kucia swobodnego modelowany jest za pomocą metody elementów skończonych i rozwiązany za pomocą programu MSC Marc. W celu przyspieszenia obliczeń zagadnienie rozważane jest w środowisku gridowym. Przedstawiono przykład numeryczny ilustrujący skuteczność zastosowanych inteligentnych technik optymalizacji.
EN
Evolutionary Computing (EC) and Ant Colony Optimization (ACO) apply stochastic searching, parallel investigation as well as autocatalitic process (or stigmergy) to solve optimization problems. This paper concentrates on the Traveling Salesman Problem (TSP) solved by evolutionary and ACO algorithms. We consider the sets of parameters and operators which influence the acting of these algorithms. Two algorithmic structures emphasizing the selection problem are discussed. We describe experiments performed for different instances of TSP problems. The comparison concludes that evolution, which is exploited especially in evolutionary algorithms, can also be observed in the performance of the ACO approach.
14
Content available remote Evolutionary computation based on Bayesian classifiers
EN
Evolutionary computation is a discipline that has been emerging for at least 40 or 50 years. All methods within this discipline are characterized by maintaining a set of possible solutions (individuals) to make them successively evolve to fitter solutions generation after generation. Examples of evolutionary computation paradigms are the broadly known Genetic Algorithms (GAs) and Estimation of Distribution Algorithms (EDAs). This paper contributes to the further development of this discipline by introducing a new evolutionary computation method based on the learning and later simulation of a Bayesian classifier in every generation. In the method we propose, at each iteration the selected group of individuals of the population is divided into different classes depending on their respective fitness value. Afterwards, a Bayesian classifier---either naive Bayes, seminaive Bayes, tree augmented naive Bayes or a similar one---is learned to model the corresponding supervised classification problem. The simulation of the latter Bayesian classifier provides individuals that form the next generation. Experimental results are presented to compare the performance of this new method with different types of EDAs and GAs. The problems chosen for this purpose are combinatorial optimization problems which are commonly used in the literature.
EN
The analysis of a population in a factor space of population states sheds light on the dynamics of reaching the equilibrium state. The evolution of states follows two phases: the fast concentration of the population and a slow movement of the almost homogenous population towards the optimum. The location of equilibrium states depends on the fitness functions. For asymmetrical fitness functions, the same phenomena as for symmetrical ones are observed: the number of fixed points depends on the modality of the fitness function, there are stable and unstable fixed points. The latter ones appear when the standard deviation of mutation was increased.
first rewind previous Strona / 1 next fast forward last
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.