8 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 8 / 8
Yayın Design and analysis of classifier learning experiments in bioinformatics: survey and case studies(IEEE Computer Soc, 2012-12) İrsoy, Ozan; Yıldız, Olcay Taner; Alpaydın, Ahmet İbrahim EthemIn many bioinformatics applications, it is important to assess and compare the performances of algorithms trained from data, to be able to draw conclusions unaffected by chance and are therefore significant. Both the design of such experiments and the analysis of the resulting data using statistical tests should be done carefully for the results to carry significance. In this paper, we first review the performance measures used in classification, the basics of experiment design and statistical tests. We then give the results of our survey over 1,500 papers published in the last two years in three bioinformatics journals (including this one). Although the basics of experiment design are well understood, such as resampling instead of using a single training set and the use of different performance metrics instead of error, only 21 percent of the papers use any statistical test for comparison. In the third part, we analyze four different scenarios which we encounter frequently in the bioinformatics literature, discussing the proper statistical methodology as well as showing an example case study for each. With the supplementary software, we hope that the guidelines we discuss will play an important role in future studies.Yayın Immitance data modelling via linear interpolation techniques: a classical circuit theory approach(Wiley-Blackwell, 2004-11) Yarman, Bekir Sıddık Binboğa; Kılınç, Ali; Aksen, AhmetWith the advancement of the manufacturing technologies to produce new generation analog/digital communication systems, immitance data modelling has gained renewed importance in the literature. Specifically, models are utilized for behaviour characterization, simulation of physical devices or to design sub-systems with active and passive solid-state devices. Therefore, in this paper, new computer aided tools are presented to model one port immitance data by means of linear interpolation techniques. The basic philosophy of the new modelling tools is based on the numerical decomposition of the immitance data into its minimum and Foster parts. Computer algorithms are presented to model the minimum and the Foster parts of the given immitance data. Implementations of these algorithms are exhibited by means of examples. Depending on the application, modelling tools based on linear interpolation techniques may present 'computational and practical' advantages over the existing interpolation techniques, non-linear curve fittings or regression methods. It is expected that the new modelling tools will be utilized to provide initial circuit topologies to the commercially available analysis/simulation and design packages.Yayın Linear expansions for frequency selective channels in OFDM(Elsevier GMBH, 2006) Şenol, Habib; Çırpan, Hakan Ali; Panayırcı, ErdalModeling the frequency selective fading channels as random processes, we employ a linear expansion based on the Karhumen-Loeve (KL) series representation involving a complete set of orthogonal deterministic vectors with a corresponding uncorrelated random coefficients. Focusing on OFDM transmissions through frequency selective fading, this paper pursues a computationally efficient, pilot-aided linear minimum mean square error (MMSE) uncorrelated KL series expansion coefficients estimation algorithm. Based on such an expansion, no matrix inversion is required in the proposed MMSE estimator. Moreover, truncation in the linear expansion of channel is achieved by exploiting the optimal truncation property of the KL expansion resulting in a smaller computational load on the estimation algorithm. The performance of the proposed approach is studied through analytical and experimental results. We first exploit the performance of the MMSE channel estimator based on the evaluation of minimum Bayesian MSE. We also provide performance analysis results studying the influence of the effect of SNR and correlation mismatch on the estimator performance. Simulation results confirm our theoretical results and illustrate that the proposed algorithm is capable of tracking fast fading and improving performance.Yayın A computer-aided design technique for lossless matching networks with mixed, lumped and distributed elements(Elsevier GMBH, 2004) Sertbaş, Ahmet; Yarman, Bekir Sıddık BinboğaA computer-aided method for the design of lossless broadband matching networks with lumped elements and commensurate transmission lines is presented. ne method is based on combining the simplifield real frequency technique with the algebraic network decomposition by Fettweis. To show the application of the Computer-Aided Design (CAD) approach, an UHF antenna matching problem is solved.Yayın Sequence estimation with transmit diversity for wireless communications(Urban & Fischer Verlag, 2003) Panayırcı, Erdal; Aygölü, Hasan Ümit; Pusane, Ali EmreIn this paper, an optimum sequence estimation algorithm for wireless systems with Alamouti's two transmitter diversity in the presence of multipath fading is proposed. The algorithm is based on a jointly iterative channel and sequence estimation according to the maximum likelihood (ML) criterion, using the Expectation-Maximization (EM) algorithm employing an M-level phase-shift keying (M-PSK) modulation scheme with additive Gaussian noise. The discrete multipath channel is represented in terms of the channel gains from each transmit antenna to the receive antenna. EM algorithm estimates jointly the complex channel parameters of each channel and the data sequence transmitted, iteratively, which converges to the true ML solution. The channel estimation is achieved in a simple way through the iterative equations by decoupling of the signals transmitted from different antennas. The algorithm is applied to the trellis coded modulation systems and the efficiency of the algorithm proposed has been shown with computer simulations. The simulation results show that the EM algorithm converges quickly for fast fading channels. The performance of the EM-based decoder approaches that of the ML receiver which has perfect knowledge of the channel.Yayın Significance map pruning and other enhancements to SPIHT image coding algorithm(Elsevier Science, 2003-10) Bayazıt, UluğThis paper proposes several enhancements to the Set Partitioning in Hierarchical Trees (SPIHT) image coding algorithm without changing the original algorithm's general skeleton. First and foremost, a method for significance map pruning based on a rate-distortion criterion is introduced. Specifically, the (Type A) sets of wavelet coefficients with small ratios of estimated distortion reduction to estimated rate contribution are deemed insignificant and effectively pruned. Even though determining such sets requires the computational complexity of the encoder to increase considerably with respect to the original SPIHT encoder, the original SPIHT decoder may still be used to decode the generated bitstream with a low computational complexity. The paper also proposes three low complexity enhancements by more sophisticated use of the adaptive arithmetic coder. Simulation results demonstrate that all these enhancements yield modest compression gains at moderate to high rates.Yayın An immitance based tool for modelling passive one-port devices by means of darlington equivalents(Urban & Fischer Verlag, 2001) Yarman, Bekir Sıddık Binboğa; Aksen, Ahmet; Kılınç, AliAn immitance-based method is presented to model measured or computed data, obtained from a "passive one-port physical device" by means of its Darlington equivalent. In other words, the given data is modelled as a lossless two port terminated in a unit resistor. The basis of the new modelling tool rests on the numerical decomposition of the given immitance data into its Foster and minimum parts. Therefore, the proposed technique does not require any choice for the circuit topology to build the model. Rather, the optimum circuit topology that characterises the given data is the natural consequence of the modelling process proposed in this paper. A main algorithm is presented to construct the model from the given data. It is expected that the proposed modelling tool will find practical applications in the behaviour characterisation, simulation, and design of high speed/high frequency analog/digital mobile communication sub-systems manufactured on VLSI chips. An antenna-modelling example is included to systematically exhibit the implementation of the modelling technique.Yayın Force-directed approaches to sensor localization(Assoc Computing Machinery, 2010-09) Efrat, Alon; Forrester, David; Iyer, Anand; Kobourov, Stephen G.; Erten, Cesim; Kılıç, Yasin OzanAs the number of applications of sensor networks increases, so does the interest in sensor network localization, that is, in recovering the correct position of each node in a network of sensors from partial connectivity information such as adjacency, range, or angle between neighboring nodes. In this article, we consider the anchor-free localization problem in sensor networks that report possibly noisy range information and angular information about the relative order of each sensor's neighbors. Previously proposed techniques seem to successfully reconstruct the original positions of the nodes for relatively small networks with nodes distributed in simple regions. However, these techniques do not scale well with network size and yield poor results with nonconvex or nonsimple underlying topology. Moreover, the distributed nature of the problem makes some of the centralized techniques inapplicable in distributed settings. To address these problems we describe a multiscale dead-reckoning (MSDR) algorithm that scales well for large networks, can reconstruct complex underlying topologies, and is resilient to noise. The MSDR algorithm takes its roots from classic force-directed graph layout computation techniques. These techniques are augmented with a multiscale extension to handle the scalability issue and with a dead-reckoning extension to overcome the problems arising with nonsimple topologies. Furthermore, we show that the distributed version of the MSDR algorithm performs as well as, if not better than, its centralized counterpart, as shown by the quality of the layout, measured in terms of the accuracy of the computed pairwise distances between sensors in the network.
