Faculty of Engineering

Permanent URI for this communityhttps://hdl.handle.net/10679/10

Browse

Now showing 1 - 20 of 3090

Metadata only
Detecting credit card fraud by modified Fisher discriminant analysis
(Elsevier, 01.04.2015) Mahmoudi, Nader; Duman, Ekrem; Industrial Engineering; DUMAN, Ekrem; Mahmoudi, Nader
In parallel to the increase in the number of credit card transactions, the financial losses due to fraud have also increased. Thus, the popularity of credit card fraud detection has been increased both for academicians and banks. Many supervised learning methods were introduced in credit card fraud literature some of which bears quite complex algorithms. As compared to complex algorithms which somehow over-fit the dataset they are built on, one can expect simpler algorithms may show a more robust performance on a range of datasets. Although, linear discriminant functions are less complex classifiers and can work on high-dimensional problems like credit card fraud detection, they did not receive considerable attention so far. This study investigates a linear discriminant, called Fisher Discriminant Function for the first time in credit card fraud detection problem. On the other hand, in this and some other domains, cost of false negatives is very higher than false positives and is different for each transaction. Thus, it is necessary to develop classification methods which are biased toward the most important instances. To cope for this, a Modified Fisher Discriminant Function is proposed in this study which makes the traditional function more sensitive to the important instances. This way, the profit that can be obtained from a fraud/legitimate classifier is maximized. Experimental results confirm that Modified Fisher Discriminant could eventuate more profit.
Metadata only
Unmanned aerial vehicle relay networks
(Help Cookies Privacy Statement Accessibility Modern Slavery Act Statement Legal notices The Institution of Engineering and Technology, 202) Adam, Evşen Yanmaz; Electrical & Electronics Engineering; Davoli, L.; Ferrari, G.; ADAM, Evşen Yanmaz
N/A
Open Access
SCARA with Path trajectory
(2008) Elamassie, Mohammed; Elamassie, Mohammed
The following Matlab project contains the source code and Matlab examples used for SCARA with Path trajectory. By defining the initial position and final position the robot will follow the path between these two points
Metadata only
Energetic reasoning revisited: application to parallel machine scheduling
(Springer Science+Business Media, 2008-08) Hidri, L.; Gharbi, A.; Haouari, Mohamed; Industrial Engineering; HAOUARI, Mohamed
We consider the problem of minimizing makespan on identical parallel machines subject to release dates and delivery times. We present several new feasibility tests and adjustment techniques that consistently improve theclassical energetic reasoning approach. Computational results carried out on a set of hard instances provide strong evidence that the performance of a state-of-the-art exact branch-and-bound algorithm is substantially improved through embedding the proposed enhanced energetic reasoning.
Metadata only
Optimal solution of the discrete cost multicommodity network design problem
(Elsevier, 2008-10-15) Mrad, M.; Haouari, Mohamed; Industrial Engineering; HAOUARI, Mohamed
We investigate a multicommodity network design problem where a discrete set of Technologies with step-increasing cost and capacity functions should be installed on the edges. This problem is a fundamental network design problem having many important applications in contemporary telecommunication networks. We describe an exact constraint generation approach and we show that the conjunctive use of valid inequalities, bipartition inequalities that are generated using max-flow computations, as well as an exact separation algorithm of metric inequalities makes it feasible to solve to optimality instances with up to 50 nodes and 100 edges.
Metadata only
Metal CMP optimization based on chemically formed thin film analysis
(The Electrochemical Society, 2009) Başım, Gül Bahar; Mechanical Engineering; BAŞIM DOĞAN, Gül Bahar
The conventional demands for development in semiconductor industry are changing as the Moore's Law is approaching to its limits. This paper demonstrates a theoretical optimization approach for the planarization of metal films by Chemical Mechanical Polishing (CMP) process. Optimal removal rate and a smooth surface finish post CMP can be achieved by the combined effect of the chemical and mechanical components of the process. Metal CMP necessitates a protective oxide film formation in the presence of surface active agents, corrosives, pH regulators etc' to achieve global planarization. Formation and mechanical properties of the chemically modified films determine the stresses develop in the film structure delineating the stability of the chemically altered films on the surface of the metal wafer. The balance between the stresses built in the film structure versus the mechanical actions provided during the process can be used to optimize the process variables and furthermore help define new planarization techniques for the next generation microelectronic device manufacturing which is expected to deal with atomic level structures.
Open Access
MathOptimizer: a nonlinear optimization package for mathematica users
(2009) Kampas, F. J.; Pinter, Janos D.; Industrial Engineering; PINTER, Janos
Mathematica is an advanced software system that enables symbolic computing, numerics, program code development, model visualization and professional documentation in a unified framework. Our MathOptimizer software package serves to solve global and local optimization models developed using Mathematica. We introduce MathOptimizer’s key features and discuss its usage options that support a range of operational modes. The numerical capabilities of the package are illustrated by simple and more advanced examples, pointing towards a broad range of potential applications.
Metadata only
Improving the performance of wireless H.264 video broadcasting through a cross-layer design
(IEEE, 2009) Atıcı, Çağdaş; Sunay, Mehmet Oğuz; Computer Science; SUNAY, Mehmet Oğuz; Atıcı, Çağdaş
In a wireless video broadcasting service, number of users receiving the service and the average video quality of the received stream have to be maximized for improving the system performance. To solve this problem optimally, one may adjust parameters at the physical and the application layers in a cross-layer fashion while utilizing the characteristics of the videothat is being transmitted. In this paper, we propose a multiobjective optimized, cross-layer video broadcasting scheme for a wireless system capable of supporting a multitude of transmission data-rates using the H.264/AVC. The multi-objective, cross-layer optimization aims to find the H.264/AVC as well as the physical layer system parameters jointly to reach the optimal compromise between maximizing the average received video PSNR and minimizing the video broadcast service outage probability. Simulations conducted for the ITU Pedestrian A and Vehicular B channels show that further gains in system performance can be achieved for video broadcasting when such a cross-layer design is used.
Metadata only
Capacity of wireless ad-hoc broadcast networks under realistic channel models
(IEEE, 2009) Atıcı, Çağdaş; Sunay, Mehmet Oğuz; Computer Science; SUNAY, Mehmet Oğuz; Atıcı, Çağdaş
In a wireless broadcasting scenario, some of the nodes can help the source node by forwarding the received information. Due to the interference from multiple transmissions, selection of these nodes directly affects the performance of the system under a given total power and hop constraint. In this paper, we first analytically find the number and the positions of the rebroadcasting nodes that achieve the optimal broadcast capacity under the continuum model. Following the results of this part, we propose two heuristics, one centralized and another distributed, for relay selection in practical scenarios. Then, we discuss the broadcast capacity performances of these algorithms under different system settings. The results illustrate that using a distributed relay selection method brings significant gains to the broadcast capacity under a realistic system model.
Metadata only
On the broadcast capacity of wireless multihop interference networks
(IEEE, 2009) Atıcı, Çağdaş; Sunay, Mehmet Oğuz; Computer Science; SUNAY, Mehmet Oğuz; Atıcı, Çağdaş
This paper is concerned with wireless broadcasting in multihop networks where a selected number of relay nodes may aid the source node in the broadcast under a given total energy and hop constraint. We study an ad-hoc network with infinitely many nodes and analytically find the number and positions of rebroadcasting relay nodes to achieve the optimal broadcast capacity. The interference due to multiple transmissions in the geographical area is taken into account. Based on the theoretical findings, we then propose one distributed and one centralized heuristic for relay selection in wireless broadcasting. We discuss the broadcast capacity performances and CSI (channel state information) requirements of these algorithms. The results illustrate that the benefits of peer-assisted broadcasting are more pronounced in the centralized relay selection algorithm when compared to the fully randomized and distributed selection under a realistic system model.
Metadata only
Improving automatic emotion recognition from speech signals
(International Speech Communications Association, 2009) Bozkurt, E.; Erzin, E.; Eroğlu Erdem, Ç.; Erdem, Tanju; Computer Science; ERDEM, Arif Tanju
We present a speech signal driven emotion recognition system. Our system is trained and tested with the INTERSPEECH 2009 Emotion Challenge corpus, which includes spontaneous and emotionally rich recordings. The challenge includes classifier and feature sub-challenges with five-class and two-class classification problems. We investigate prosody related, spectral and HMM-based features for the evaluation of emotion recognition with Gaussian mixture model (GMM) based classifiers. Spectral features consist of mel-scale cepstral coefficients (MFCC), line spectral frequency (LSF) features and their derivatives, whereas prosody-related features consist of mean normalized values of pitch, first derivative of pitch and intensity. Unsupervised training of HMM structures are employed to define prosody related temporal features for the emotion recognition problem. We also investigate data fusion of different features and decision fusion of different classifiers, which are not well studied for emotion recognition framework. Experimental results of automatic emotion recognition with the INTERSPEECH 2009 Emotion Challenge corpus are presented.
Metadata only
Approximation algorithms for single machine scheduling with one unavailability period
(Springer Nature, 2009-03) Kacem, I.; Haouari, Mohamed; Industrial Engineering; HAOUARI, Mohamed
In this paper, we investigate the single machine scheduling problem with release dates and tails and a planned unavailability time period. We show that the problem admits a fully polynomial-time approximation scheme when the tails are equal. We derive an approximation algorithm for the general case and we show that the worst-case bound of the sequence yielded by Schrage’s algorithm is equal to 2 and that this bound is tight. Some consequences of this result are also presented.
Open Access
Pseudoscalar-meson-octet-baryon coupling constants in two-flavor lattice QCD
(American Physical Society, 2009-04-01) Erkol, Güray; Oka, M.; Takahashi, T. T.; Natural and Mathematical Sciences; ERKOL, Güray
We evaluate the πΝΝ, π∑∑, πΛ∑, KΛN and K∑N coupling constants and the corresponding monopole masses in lattice QCD with two flavors of dynamical quarks. The parameters representing the SU(3)-flavor symmetry are computed at the point where the three quark flavors are degenerate at the physical s-quark mass. In particular, we obtain α ≡ F/(F+D) = 0.395(6). The quark-mass dependences of the coupling constants are obtained by changing the u- and the d-quark masses. We find that the SU(3)-flavor parameters have weak quark-mass dependence and thus the SU(3)-flavor symmetry is brokenby only a few percent at each quark-mass point we consider.
Open Access
Spin content of Λ in QCD sum rules
(American Physical Society, 2009-06-01) Erkol, Güray; Oka, M.; Natural and Mathematical Sciences; ERKOL, Güray
We calculate the isoscalar axial-vector coupling constants of the Λ hyperon using the method of QCD sum rules. A determination of these coupling constants reveals the individual contributions of the u, d, and the s quarks to the spin content of Λ. Our results for the light-quark contributions are in agreement with those from experiment assuming flavor SU(3). We also find that the flavor-SU(3)-breaking effects are small and the contributions from the u and the d quarks to the Λ polarization are negatively polarized as in the flavor-SU(3) limit.
Metadata only
Mechanical and wear properties of ultrafine-grained pure Ti produced by multi-pass equal-channel angular extrusion
(Elsevier, 2009-08-20) Purcek, G.; Saray, O.; Kul, O.; Yapıcı, Güney Güven; Haouaoui, M.; Maier, H. J.; Mechanical Engineering; YAPICI, Güney Güven
In this study, pure grade 2 Ti was processed via equal-channel angular extrusion (ECAE) for 8 and 12 passes following route-E at 300 °C. After processing, the microstructural evolution, tensile properties and wear behavior were investigated. ECAE-processed Ti exhibited a significant improvement in strength values with a slight decrease in ductility. However, the wear test results surprisingly showed that the strengthening of titanium by ECAE processing does not lead to the improvement of wear resistance at least for the pressures and sliding distances used in this study. This finding was mainly attributed to the tribochemical reaction leading to oxidative wear with the abrasive effect in Ti. Three distinct regions were formed on the subsurface of CG and UFG Ti after sliding wear, which are the tribolayer including titanium oxide with smeared wear material at the top, a deformed region having material structure oriented along the sliding direction in the middle, and the original unaffected bulk material at the bottom.
Metadata only
Heuristics for the variable sized bin-packing problem
(Elsevier, 2009-10) Haouari, Mohamed; Serairi, M.; Industrial Engineering; HAOUARI, Mohamed
We investigate the one-dimensional variable-sized bin-packing problem. This problem requires packing a set of items into a minimum-cost set of bins of unequal sizes and costs. Six optimization-based heuristics for this problem are presented and compared. We analyze their empirical performance on a large set of randomly generated test instances with up to 2000 items and seven bin types. The first contribution of this paper is to provide evidence that a set covering heuristic proves to be highly effective and capable of delivering very-high quality solutions within short CPU times. In addition, we found that a simple subset-sum problem-based heuristic consistently outperforms heuristics from the literature while requir- ing extremely short CPU times.
Metadata only
Solution of near-field thermal radiation in one-dimensional layered media using dyadic Green's functions and the scattering matrix method
(Elsevier, 2009-12) Francoeur, M.; Mengüç, Mustafa Pınar; Vaillon, R.; Mechanical Engineering; MENGÜÇ, Mustafa Pınar
A general algorithm is introduced for the analysis of near-field radiative heat transfer in one-dimensional multi-layered structures. The method is based on the solution of dyadic Green's functions, where the amplitude of the fields in each layer is calculated via a scattering matrix approach. Several tests are presented where cubic boron nitride is used in the simulations. It is shown that a film emitter thicker than 1 μm provides the same spectral distribution of near-field radiative flux as obtained from a bulk emitter. Further simulations have pointed out that the presence of a body in close proximity to an emitter can alter the near-field spectrum emitted. This algorithm can be employed to study thermal one-dimensional layered media and photonic crystals in the near-field in order to design radiators optimizing the performances of nanoscale-gap thermophotovoltaic power generators.
Metadata only
Performance analysis of cooperative diversity networks with imperfect channel estimation over Rician fading channels
(IEEE, 2010) Ikki, S. S.; Feteiha, M.; Uysal, Murat; Electrical & Electronics Engineering; UYSAL, Murat
In this paper, we examine the effect of channel estimation error on the error and outage probabilities of a multi-relay system with amplify-and-forward relaying over frequency-flat Rician fading channel. We consider orthogonal relaying and study both conventional cooperative systems (i.e., all relays participate in the relaying phase) and opportunistic cooperative systems (i.e., only the best relay participates in the relaying phase). Based on the derivation of effective signal-to-ratio (SNR) at the receiver taking into account channel estimation error, we obtain closed-form expressions for error and outage probabilities in high SNR regime. Such closed form solutions are highly desirable because they allow for rapid and efficient evaluation of system performance. Computer simulations are used to validate our analytical results.
Metadata only
Gauss karışım modeli tabanlı konuşmacı belirleme sistemlerinde klasik MAP uyarlanması yönteminin performans analizi
(IEEE, 2010) Erdoğan, A.; Demiroğlu, Cenk; Electrical & Electronics Engineering; DEMİROĞLU, Cenk
Gaussian mixture models (GMM) is one of the most commonly used methods in text-independent speaker identification systems. In this paper, performance of the GMM approach has been measured with different parameters and settings. Voice activity detection (VAD) component has been found to have a significant impact on the performance. Therefore, VAD algorithms that are robust to background noise have been proposed. Significant differences in performance have been observed between male and female speakers and GSM/PSTN channels. Moreover, single-stream GMM approach has been found to perform significantly better than the multi-stream GMM approach. It has been observed under all conditions that data duration is critical for good performance.
Metadata only
Algorithmic expedients for the prize collecting Steiner tree problem
(Elsevier, 2010) Haouari, Mohamed; Layeb, S. B.; Sherali, H. D.; Industrial Engineering; HAOUARI, Mohamed
This paper investigates the Prize Collecting Steiner Tree Problem (PCSTP) on a graph, which is a generalization of the well-known Steiner tree problem. Given a root node, edge costs, node prizes and penalties, as well as a preset quota, the PCSTP seeks to find a subtree that includes the root node and collects a total prize not smaller than the specified quota, while minimizing the sum of the total edge costs of the tree plus the penalties associated with the nodes that are not included in the subtree. For this challenging network design problem that arises in telecommunication settings, we present two valid 0-1 programming formulations and use them to develop preprocessing procedures for reducing the graph size. Also, we design an optimization-based heuristic that requires solving a PCSTP on a specific tree-subgraph. Although, this latter special case is shown to be NP-hard, it is effectively solvable in pseudo-polynomial time. The worst-case performance of the proposed heuristic is also investigated. In addition, we describe new valid inequalities for the PCSTP and embed all the aforementioned constructs in an exact row-generation approach. Our computational study reveals that the proposed approach can solve relatively large-scale PCSTP instances having up to 1000 nodes to optimality.

Browse

Browsing by Issue Date