• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 76
  • 22
  • 14
  • 13
  • 12
  • 6
  • 3
  • 2
  • 1
  • 1
  • 1
  • Tagged with
  • 182
  • 47
  • 29
  • 23
  • 22
  • 22
  • 21
  • 21
  • 19
  • 19
  • 17
  • 15
  • 15
  • 14
  • 12
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
31

Efficient algorithms for discrete geometry problems / Efikasni algoritmi za probleme iz diskretne geometrije

Savić Marko 25 October 2018 (has links)
<p>The first class of problem we study deals with geometric matchings. Given a set<br />of points in the plane, we study perfect matchings of those points by straight line<br />segments so that the segments do not cross. Bottleneck matching is such a matching that minimizes the length of the longest segment. We are interested in finding a bottleneck matching of points in convex position. In the monochromatic case, where any two points are allowed to be matched, we give an O(n <sup>2 </sup>)-time algorithm for finding a bottleneck matching, improving upon previously best known algorithm of O(n <sup>3 </sup>) time complexity. We also study a bichromatic version of this problem, where each point is colored either red or blue, and only points of different color can be matched. We develop a range of tools, for dealing with bichromatic non-crossing matchings of points in convex<br />position. Combining that set of tools with a geometric analysis enable us to solve the<br />problem of finding a bottleneck matching in O(n <sup>2 </sup>) time. We also design an O(n)-time<br />algorithm for the case where the given points lie on a circle. Previously best known results were O(n 3 ) for points in convex position, and O(n log n) for points on a circle.<br />The second class of problems we study deals with dilation of geometric networks.<br />Given a polygon representing a network, and a point p in the same plane, we aim to<br />extend the network by inserting a line segment, called a feed-link, which connects<br />p to the boundary of the polygon. Once a feed link is fixed, the geometric dilation<br />of some point q on the boundary is the ratio between the length of the shortest path<br />from p to q through the extended network, and their Euclidean distance. The utility of<br />a feed-link is inversely proportional to the maximal dilation over all boundary points.<br />We give a linear time algorithm for computing the feed-link with the minimum overall<br />dilation, thus improving upon the previously known algorithm of complexity that is<br />roughly O(n log n).</p> / <p>Prva klasa problema koju proučavamo tičee se geometrijskih mečinga. Za dat skup tačaaka u ravni, posmatramo savr&scaron;ene mečinge tih tačaka spajajućii ih&nbsp; dužima koje &nbsp; se ne smeju sećui. Bottleneck mečing je takav mečing koji minimizuje dužinu najduže duži. Na&scaron; cilj je da nađemo bottleneck mečiing tačaka u konveksnom položaju.Za monohromatski slučaj, u kom je dozvoljeno upariti svaki par tačaka, dajemo algoritam vremenske složenosti O(n <sup>2</sup>) za nalaženje bottleneck mečinga. Ovo&nbsp; je bolje od prethodno najbolji poznatog algoritam, čiija je složenost O(n <sup>3 </sup>). Takođe proučavamo bihromatsku verziju ovog problema, u kojoj je svaka tačka&nbsp; obojena ili u crveno ili u plavo, i dozvoljeno je upariti samo tačke različite boje. Razvijamo niz alata za rad sa bihromatskim nepresecajućim mečinzima tačaka u konveksnom položaju. Kombinovanje ovih alata sa geometrijskom analizom omogućava nam da re&scaron;imo problem nalaženja bottleneck mečinga u O(n <sup>2</sup> ) vremenu. Takođe, konstrui&scaron;emo algoritam vremenske složenosti O(n) za slučaj kada&nbsp; sve date tačkke leže na krugu. Prethodno najbolji poznati algoritmi su imali složenosti&nbsp; O(n <sup>3</sup> ) za tačkeke u konveksnom položaju i O(n log n) za tačke na krugu.<br />Druga klasa problema koju proučaavamo tiče se dilacije u geometrijskim mrežama. Za datu mrežu predstavljenu poligonom, i tačku p u istoj ravni, želimo pro&scaron;iriti mrežu&nbsp; dodavanjem duži zvane feed-link koja povezuje p sa obodom poligona. Kada je feed- link fiksiran, defini&scaron;emo geometrijsku dilaciju neke tačke q na obodu kao odnos izme&nbsp; đu&nbsp; dužine najkraćeg puta od p do q kroz pro&scaron;irenu mrežu i njihovog Euklidskog rastojanja. Korisnost feed-linka je obrnuto proporcionalna najvećoj dilaciji od svih ta čaka na obodu poligona. Konstrui&scaron;emo algoritam linearne vremenske složenosti koji nalazi feed-link sa najmanom sveukupnom dilacijom. Ovim postižemo bolji rezultat od prethodno najboljeg poznatog algoritma složenosti približno O(n log n).</p>
32

Efficientnext: Efficientnet For Embedded Systems

Deokar, Abhishek 05 1900 (has links)
Indiana University-Purdue University Indianapolis (IUPUI) / Convolutional Neural Networks have come a long way since AlexNet. Each year the limits of the state of the art are being pushed to new levels. EfficientNet pushed the performance metrics to a new high and EfficientNetV2 even more so. Even so, architectures for mobile applications can benefit from improved accuracy and reduced model footprint. The classic Inverted Residual block has been the foundation upon which most mobile networks seek to improve. EfficientNet architecture is built using the same Inverted Residual block. In this thesis we experiment with Harmonious Bottlenecks in place of the Inverted Residuals to observe a reduction in the number of parameters and improvement in accuracy. The designed network is then deployed on the NXP i.MX 8M Mini board for Image classification. / 2023-10-11
33

Do Severe Genetic Bottlenecks Lead to Greater Reproductive Failure?

Burrows, Ben Robert January 2006 (has links)
It is generally accepted that populations which experience severe bottlenecks have a reduction in fitness. One of the most frequently reported fitness costs is increased hatching failure in bottlenecked populations of birds. The mechanism responsible for increased hatching failure is unknown. Research on other animals suggest that reduced population numbers cause unavoidable inbreeding that in turn leads to abnormalities in the gametes. In this thesis I examine some of the possible causes for increased hatching failure in severely bottlenecked populations of introduced birds in New Zealand. I look at three traits identified as a cause for infertility or hatching failure previously and determine whether there is a link with the size of a population s bottleneck. It is possible that reduced numbers of sperm reaching the site of fertilisation is a primary cause of hatching failure. I examined the perivitelline membrane of various species of introduced birds and counted the total number of sperm present to compare to how many would be expected in non-bottlenecked species. Although there was no relationship between the size of the bottleneck and the number of sperm present, all species had lower than expected sperm counts. In many species of mammals, a reduction in the quality of sperm is attributed to inbreeding depression bought about by genetic bottlenecks. I next compared the level of sperm abnormalities, variation in midpiece size sperm, and sperm motility with the size of the bottleneck each species passed through when introduced to New Zealand. There was no significant correlation between either the variation in midpiece size or sperm motility with bottleneck size. However, there was a trend for species that passed through more severe bottlenecks to have a slightly higher level of midpiece size and lower motility. Finally, I examined whether there was a link between abnormalities in the eggshell and the size of the respective bottleneck. There was no significant change in eggshell thickness or any change in the number of pores associated bottleneck size. However, there was a decreased number of round pores in severely bottlenecked species, although the consequences of this are unknown. My findings do not directly link a single cause for increased hatching failure in bottlenecked species of birds, but they do highlight the need for monitoring of reproductive traits in endangered species that have experienced a population bottleneck.
34

Model dopravního toku s překážkou / A traffic flow with a bottelneck

Kovařík, Adam January 2011 (has links)
Title: A traffic flow with a bottelneck Author: Adam Kovařík Department: Department of Numerical Mathematics Supervisor: prof. RNDr. Vladimír Janovský, DrSc. Supervisor's e-mail address: janovsky@karlin.mff.cuni.cz Abstract: In this paper we study a microscopic follow-the-leader traffic model on a circu- lar road with a bottleneck. We assume that all drivers are identical and overtaking is not permitted. We sketch a small part of the rich dynamics of the model including Hopf and Neimark-Sacker bifurcations. We introduce so called POM and quasi-POM solutions and an algorithm how to search them. The main goal of this work is to investigate how the optimal velocity model with a bottleneck deals with so called aggressive behavior of dri- vers. The effect of variable reaction time and a combination of both named factors is also tested. Using numerical simulations we'll find out that aggressiveness and faster reactions have positive effect on traffic flow. In the end we discuss models with two bottlenecks and with one extraordinary driver. Keywords: dynamical systems, ODEs, traffic flow, bottleneck, aggressiveness. 1
35

Acesso paralelo a arquivos em sistemas multiprocessadores. / Parallel access to files for multiprocessor systems.

Vasata, Darlon 12 February 2010 (has links)
Este trabalho apresenta um estudo sobre o gargalo no acesso a disco sob arquiteturas com sistemas multiprocessadores e processadores multicore e apresenta uma proposta de solução para o problema e sua implementação. A solução desenvolvida é chamada de FPA. A estratégia desenvolvida no trabalho aproveita os recursos de memória compartilhada da arquitetura e utiliza buffers de dados compartilhados entre as diversas threads das aplicações, bem como tenta otimizar o acesso ao disco rígido. Para que seja diminuída a concorrência no acesso a disco, é permitido que apenas uma thread realize leituras ou gravações no disco por vez, e nas situações de escrita é feita a sobreposição entre o processamento e o envio dos dados ao disco. As operações de escrita são realizadas por uma thread adicional, que é inicializada e gerenciada pela FPA. Foram executados diversos testes com diferentes números de threads, tamanhos de arquivo e quantidade de processamento de cálculos na aplicação. Os resultados apresentaram ganho de desempenho com a FPA em diversas situações, onde um dos fatores mais significativos para o ganho de desempenho é a quantidade de processamento existente na aplicação. De modo geral, as operações de leitura apresentaram ganho de desempenho em relação ao uso de streams nas situações onde existe pouco processamento de cálculos na aplicação, e as operações de escrita mostraram vantagem nas situações onde existe uma quantidade de processamento de cálculos maior na aplicação. Nas operações de escrita é obtido vantagem com a sobreposição entre processamento e escrita de dados. Porém, quando a quantidade de processamento de cálculos é muito alta, o tempo de espera pela gravação em disco torna-se irrisório perante o tempo gasto no processamento dos cálculos, ocasionando perda de desempenho da FPA em relação aos streams. / This work presents a study on disk access bottlenecks for multiprocessor architecture systems and also proposes a specification to solve this problem and its implementation. The developed strategy is called FPA. The developed strategy makes use of shared memory architecture and uses shared data buffers among several application threads and also tries to optimize disk access time. In order to decrease disk concurrency, is allowed to just one application thread to read/write simultaneously and processing thread operations run parallelly to disk operations. Write operations are performed by an internal thread, which is started and managed by FPA. Several tests were performed concerning different amounts of application threads instances, file sizes and computing instructions. Results show performance gain with FPA in several situations. The main factor that affects performance is the amount of computing instructions on each application. In general, read operations showed a better performance compared to use of streams in situations where there is few computing instructions in the application, and write operations showed advantages on situations where there are more computing instructions in the application. Write operations are advantageous using data writing and computing instructions overlap. However, in cases of a large amount of computing instructions compared to disk operations, the time to perform disk operations becomes too few compared to computing time, occurring FPA performance loss compared to streams.
36

Método estruturado para aplicação das técnicas de aumento da capacidade de produção de recursos gargalo em células de manufatura / Structured method for the application of techniques to increase bottleneck resource production capacity in manufacturing cells

Martins Junior, José Carlos 31 August 2009 (has links)
O propósito deste trabalho é estruturar uma sequência de aplicação de uma série de técnicas de aumento da capacidade de produção em um equipamento gargalo citadas na literatura. Esta sequência deverá ser colocada em prática para aumentar a capacidade de produção de linhas de fabricação em ambientes de manufatura enxuta. O trabalho foi desenvolvido através da análise das técnicas encontradas para aumento de capacidade de produção de equipamentos (Pesquisa Aplicada) e da estruturação de uma aplicação seqüencial das técnicas (Pesquisa-Ação). Esta sequência de aplicação é seguida até que a melhoria de capacidade de produção necessária seja atingida. A seqüência de aplicação desenvolvida foi aplicada em uma linha de arranjo celular numa empresa de autopeças. As metodologias de pesquisa utilizadas foram a pesquisa aplicada e pesquisa-ação. Foi possível observar que os assuntos que mais contribuíram para as bases deste trabalho foram a Teoria das Restrições e a Produção Enxuta, em particular o processo de focalização em cinco etapas da Teoria das Restrições. Foi possível identificar a necessidade de aumento da capacidade de produção tanto quando a capacidade é insuficiente para atendimento do cliente como quando uma estratégia de redução de custo impõe uma impossibilidade de atendimento à demanda do cliente. Futuros trabalhos podem explorar as diferentes incidências de aplicação de cada uma das técnicas apresentadas para cada tipo de equipamento. Tipos estes ligados à indústria, complexidade, porte, custo, etc. Novas técnicas podem surgir e ser anexadas à seqüência apresentada. Esta forma estruturada de elevação da restrição, neste caso o gargalo de fabricação, poderá ser utilizada em diferentes ambientes fabris pelos profissionais responsáveis por estas ações de melhoria. / The purpose of this work is structuring a sequence of application of a series of production capacity increase techniques mentioned in literature on an equipment bottleneck. This sequence must be placed into practice to increase production capacity of manufacturing lines in lean manufacturing environments. The work was developed through the analysis of the techniques have been found to increase equipment production capacity (applied research) and a sequential procedure to techniques application (action research). This sequential procedure is followed up until the improvement required production capacity is reached. The sequential procedure developed was applied in one line of cellular arrangement in an auto parts company. Research methodologies used were the applied research and research-action. It was possible to observe that matters most contributed to the foundations of this work were the theory of constraints and lean production, in particular the process of focus on five stages of the theory of constraints. It was possible to identify the need of using this sequential procedure when production capacity is insufficient to support the customer demand as when a cost reduction strategy imposes an inability to support the customer demand. Future work can explore the different number of using of each of the techniques presented for each type of equipment. These equipment types are related to complexity, cost, etc. New techniques may arise and be attached to the sequential procedure. This structured way of elevating the restriction in this case the bottleneck manufacturing could be used in different environments works by professionals responsible for these actions for improvement.
37

Acesso paralelo a arquivos em sistemas multiprocessadores. / Parallel access to files for multiprocessor systems.

Darlon Vasata 12 February 2010 (has links)
Este trabalho apresenta um estudo sobre o gargalo no acesso a disco sob arquiteturas com sistemas multiprocessadores e processadores multicore e apresenta uma proposta de solução para o problema e sua implementação. A solução desenvolvida é chamada de FPA. A estratégia desenvolvida no trabalho aproveita os recursos de memória compartilhada da arquitetura e utiliza buffers de dados compartilhados entre as diversas threads das aplicações, bem como tenta otimizar o acesso ao disco rígido. Para que seja diminuída a concorrência no acesso a disco, é permitido que apenas uma thread realize leituras ou gravações no disco por vez, e nas situações de escrita é feita a sobreposição entre o processamento e o envio dos dados ao disco. As operações de escrita são realizadas por uma thread adicional, que é inicializada e gerenciada pela FPA. Foram executados diversos testes com diferentes números de threads, tamanhos de arquivo e quantidade de processamento de cálculos na aplicação. Os resultados apresentaram ganho de desempenho com a FPA em diversas situações, onde um dos fatores mais significativos para o ganho de desempenho é a quantidade de processamento existente na aplicação. De modo geral, as operações de leitura apresentaram ganho de desempenho em relação ao uso de streams nas situações onde existe pouco processamento de cálculos na aplicação, e as operações de escrita mostraram vantagem nas situações onde existe uma quantidade de processamento de cálculos maior na aplicação. Nas operações de escrita é obtido vantagem com a sobreposição entre processamento e escrita de dados. Porém, quando a quantidade de processamento de cálculos é muito alta, o tempo de espera pela gravação em disco torna-se irrisório perante o tempo gasto no processamento dos cálculos, ocasionando perda de desempenho da FPA em relação aos streams. / This work presents a study on disk access bottlenecks for multiprocessor architecture systems and also proposes a specification to solve this problem and its implementation. The developed strategy is called FPA. The developed strategy makes use of shared memory architecture and uses shared data buffers among several application threads and also tries to optimize disk access time. In order to decrease disk concurrency, is allowed to just one application thread to read/write simultaneously and processing thread operations run parallelly to disk operations. Write operations are performed by an internal thread, which is started and managed by FPA. Several tests were performed concerning different amounts of application threads instances, file sizes and computing instructions. Results show performance gain with FPA in several situations. The main factor that affects performance is the amount of computing instructions on each application. In general, read operations showed a better performance compared to use of streams in situations where there is few computing instructions in the application, and write operations showed advantages on situations where there are more computing instructions in the application. Write operations are advantageous using data writing and computing instructions overlap. However, in cases of a large amount of computing instructions compared to disk operations, the time to perform disk operations becomes too few compared to computing time, occurring FPA performance loss compared to streams.
38

Conservation Genetics of the White-Tailed Eagle

Hailer, Frank January 2006 (has links)
<p>The white-tailed eagle is a formerly threatened raptor that is commonly used as a flagship and indicator species in conservation work. This thesis uses molecular genetic methods to study sex determination of nestlings, genetic variability, population structure and phylogeography of the white-tailed eagle.</p><p>Fourteen microsatellite markers were developed and tested for the white-tailed eagle.</p><p>A method to sex white-tailed eagle nestlings in the field is presented. The method is based on just one tarsus measure, and is suitable for situations where a single person is handling the nestlings alone in a treetop.</p><p>Most European white-tailed eagle populations underwent extreme declines during the 20th century. The results presented here show that bottlenecked populations have maintained significant levels of genetic diversity. Gene flow between regions is not a main explanation for this, as indicated by both genetic and ringing data. Instead, the long generation time of white-tailed eagles has acted as an intrinsic buffer against rapid loss of genetic diversity. Additionally, local conservation led to protection of more genetic diversity than if conservation had focused on the large remnant population in Norway.</p><p>Mitochondrial DNA of white-tailed eagles is structured in two main clades with a predominantly eastern and western Eurasian distribution. The clades likely correspond to separate Ice Age refugia but do not grant classification as evolutionary significant units given their current extensive overlap across large parts of Eurasia.</p><p>Microsatellite variation was studied in populations across Eurasia. Variability was rather constant across the continent, but clearly lower on Iceland and Greenland. This is best explained by founder effects during their colonisation, but only weak bottlenecks during colonisation of and persistence on the continent. Current population differentiation between Europe and eastern Eurasia is not compatible with a zero gene flow model but requires some amount of gene flow over evolutionary time scales.</p>
39

Hierarchical Multi-Bottleneck Classification Method And Its Application to DNA Microarray Expression Data

Xiong, Xuejian, Wong, Weng Fai, Hsu, Wen Jing 01 1900 (has links)
The recent development of DNA microarray technology is creating a wealth of gene expression data. Typically these datasets have high dimensionality and a lot of varieties. Analysis of DNA microarray expression data is a fast growing research area that interfaces various disciplines such as biology, biochemistry, computer science and statistics. It is concluded that clustering and classification techniques can be successfully employed to group genes based on the similarity of their expression patterns. In this paper, a hierarchical multi-bottleneck classification method is proposed, and it is applied to classify a publicly available gene microarray expression data of budding yeast Saccharomyces cerevisiae. / Singapore-MIT Alliance (SMA)
40

Sensitivity Analysis for Shortest Path Problems and Maximum Capacity Path Problems in Undirected Graphs

Ramaswamy, Ramkumar, Orlin, James B., Chakravarty, Nilopal 30 April 2004 (has links)
This paper addresses sensitivity analysis questions concerning the shortest path problem and the maximum capacity path problem in an undirected network. For both problems, we determine the maximum and minimum weights that each edge can have so that a given path remains optimal. For both problems, we show how to determine these maximum and minimum values for all edges in O(m + K log K) time, where m is the number of edges in the network, and K is the number of edges on the given optimal path.

Page generated in 0.0402 seconds