Scholarly Publications - Electrical and Electronics Engineering

Permanent URI for this collectionhttps://hdl.handle.net/11693/115599

Browse

Now showing 1 - 20 of 4301

Open Access
Minimizing electric fields and increasing peripheral nervestimulation thresholds using a body gradient array coil
(John Wiley & Sons, Inc., 2024-04-16) Babaloo, Reza; Atalar, Ergin
**Purpose:** To demonstrate the performance of gradient array coils in minimizing switched-gradient-induced electric fields (E-fields) and improving peripheral nerve stimulation (PNS) thresholds while generating gradient fields with adjustable linearity across customizable regions of linearity (ROLs). **Methods:** A body gradient array coil is used to reduce the induced E-fields on the surface of a body model by modulating applied currents. This is achieved by performing an optimization problem with the peak E-field as the objective function and current amplitudes as unknown variables. Coil dimensions and winding patterns are fixed throughout the optimization, whereas other engineering metrics remain adjustable. Various scenarios are explored by manipulating adjustable parameters. **Results:** The array design consistently yields lower E-fields and higher PNS thresholds across all scenarios compared with a conventional coil. When the gradient array coil generates target gradient fields within a 44-cm-diameter spherical ROL, the maximum E-field is reduced by 10%, 18%, and 61% for the X, Y, and Z gradients, respectively. Transitioning to a smaller ROL (24 cm) and relaxing the gradient linearity error results in further E-field reductions. In oblique gradients, the array coil demonstrates the most substantial reduction of 40% in the Z–Y direction. Among the investigated scenarios, the most significant increase of 4.3-fold is observed in the PNS thresholds. **Conclusion:** Our study demonstrated that gradient array coils offer a promising pathway toward achieving high-performance gradient coils regarding gradient strength, slew rate, and PNS thresholds, especially in scenarios in which linear magnetic fields are required within specific target regions.
Open Access
Foreword
(2024) Aydın, Orhan; Özen, Ali; Köse, Ercan; Çiçek, Serdar; Biçer, Mustafa Berkan; Alcan, Veysel; Obuz, Serhat; Üstün, Deniz; Şenol, Ali; Ateş, Volkan; Güven, Sara Altın; Develi, İbrahim; Akarun, Lale; Yazar, Ahmet; Ünverdi, Özlem; Çekiç, Yalçın; Yeğin, Alper; Acar, Yusuf; Çetin, A. Enis; Erdem, Erkut; Altuncu, Ahmet; Özgür, Arzucan; Yılmaz, Atila; Akan, Aydın; Töreyin, Behçet Uğur; Tekin, Cem; Akbaş, Emre; Aktaş, Emre; Ekşioğlu, Ender Mete; Özkaynak, Fatih; Akar, Gözde Bozdağı; Keleş, Hacer Yalım; Çevikalp, Hakan; Ateş, Hasan F.; Ertuğrul, Itır Önal; Kaya, İsmail; Özkan, Kemal; Çağıltay, Kürşat; Göktürk, Mehmet; Palandöken, Merih; Cinbiş, Nazlı İkizler; Taşpınar, Necmi; Kucur, Oğuz; Kızılbey, Oğuzhan; Erkent, Özgür; Ertuğ, Özgür; İncel, Özlem Durmaz; Duygulu, Pınar; Sarıtaş, Serkan; Aydemir, Sibel; Çolak, Sultan Aldırmaz; Kayıkçıoğlu, Temel; Özkurt, Tolga Esat; Genç, Yakup; Özkazanç, Yakup
Open Access
Actor prioritized experience replay (abstract reprint)
(AAAI Press, 2024-03-24) Sağlam, Baturay; Mutlu, Furkan; Çiçek, Doğan; Kozat, Süleyman
A widely-studied deep reinforcement learning (RL) technique known as Prioritized Experience Replay (PER) allows agents to learn from transitions sampled with non-uniform probability proportional to their temporal-difference (TD) error. Although it has been shown that PER is one of the most crucial components for the overall performance of deep RL methods in discrete action domains, many empirical studies indicate that it considerably underperforms off-policy actor-critic algorithms. We theoretically show that actor networks cannot be effectively trained with transitions that have large TD errors. As a result, the approximate policy gradient computed under the Q-network diverges from the actual gradient computed under the optimal Q-function. Motivated by this, we introduce a novel experience replay sampling framework for actor-critic methods, which also regards issues with stability and recent findings behind the poor empirical performance of PER. The introduced algorithm suggests a new branch of improvements to PER and schedules effective and efficient training for both actor and critic networks. An extensive set of experiments verifies our theoretical findings, showing that our method outperforms competing approaches and achieves state-of-the-art results over the standard off-policy actor-critic algorithms.
Open Access
Learning the pareto set under incomplete preferences: pure exploration in vector bandits
(2024-11-26) Karagözlü, Efe Mert; Yıldırım, Yaşar Cahit; Ararat, Çagın; Tekin, Cem; Dasgupta, S; Mandt, S; Li, Y
We study pure exploration in bandit problems with vector-valued rewards, where the goal is to (approximately) identify the Pareto set of arms given incomplete preferences induced by a polyhedral convex cone. We address the open problem of designing sampleefficient learning algorithms for such problems. We propose Pareto Vector Bandits (PaVeBa), an adaptive elimination algorithm that nearly matches the gap-dependent and worst-case lower bounds on the sample complexity of (., d)-PAC Pareto set identification. Finally, we provide an in-depth numerical investigation of PaVeBa and its heuristic vari-ants by comparing them with the state-of-the-art multi-objective and vector optimization algorithms on several real-world datasets with conflicting objectives.
Open Access
Automatic construction of sememe knowledge bases from machine readable dictionaries
(Institute of Electrical and Electronics Engineers, 2023-12-28) Battal, Ömer Musa; Koç, Aykut
Sememes are the minimum semantic units of natural languages. Words annotated with sememes are organized into Sememe Knowledge Bases (SKBs). SKBs are successfully applied to various high-level language processing tasks as external knowledge bases. However, existing SKBs are manually or semi-manually constructed by linguistic experts over long periods, inhibiting their widespread utilization, updating, and expansion. To automatically construct an SKB from Machine-Readable Dictionaries (MRDs), which are readily available, we propose MRD2SKB as an automatic SKB generation approach. Well-established MRDs exist, and their construction is much simpler than SKBs. Therefore, the proposed MRD2SKB allows for fast, flexible, and extendable generation of SKBs. Building upon matrix factorization and topic modeling, we proposed several variants of MRD2SKB and constructed SKBs fully automatically. Both quantitative and qualitative results of extensive experiments are presented to demonstrate that the performances of the proposed automatically created SKBs are on par with manually and semi-manually prepared SKBs.
Open Access
Investigation of ballistic behavior of aluminum foam
(Springer Singapore, 2024-10-27) Kuşhan, Melih Cemal; Daz, Batuhan; Ünalır, Tolga; Çetin, Barış; Göde, Engin; Kuşhan, Canatay Battal; Tonbul, Kürşat; Gürü, Metin; Yang, Huachao; Wong, Kok Hoe
In this paper, preliminary results of the use of closed-cell aluminum foams with different densities, an innovative material, in ceramic-faced armor systems are presented. Within the scope of this study, aluminum foams of different densities were sandwiched with a well-pressed UHMWPE (Ultra High Molecular Weight Polyethylene) composite plate and used as a backing plate in the ceramic-faced armor system, and the effect of density of closed-cell aluminum foams was investigated. In the armor design in this study; Polyurea was used as the encapsulation polymer, alumina with high compressive strength and hardness to deform ammunition, UHWMPE to absorb energy and closed-cell aluminum foam to absorb energy were used. Within the scope of this paper; ballistic tests were performed on armor samples prepared with three different densities (0.30 g/cm3, 0.38 g/cm3 and 0.50 g/cm3) of aluminum foam against 7.62 mm x 63 M2 AP ammunition defined at Level 4 in NIJ 0108.01 standard and the test results were presented. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.
Unknown
Kernel-based fast factorization techniques
(Institution of Engineering and Technology, 2024-08-06) Ergül, Özgür; Khalichi, Bahram; Ertürk, Vakur B.
This chapter has focused on MLFMA as a representative kernel-based fast factorization technique. To construct a basis for further discussion, we first considered the conventional MLFMA, which is based on the plane-wave expansion of electromagnetic waves, at a formulation level. To solve multi-scale problems involving dense (uniform or non-uniform) discretizations of electrically large objects, alternative MLFMA versions are needed since the conventional MLFMA suffers from a low-frequency breakdown. We listed a variety of ways to implement low-frequency-stable MLFMAs, such as based on multipoles, inhomogeneous plane waves, coordinate shifts, and approximation techniques. We showed how MLFMA implementations can be used to solve extremely large problems via parallelization, while they can be applied to complex structures with different material properties, including plasmonic and NZI objects. Examples were given for solutions of densely discretized objects to demonstrate how MLFMA can handle such complicated problems that possess modeling challenges. Finally, problems with non-uniform discretizations that naturally arise in multi-scale simulations were considered. A rigorous implementation for stable, accurate, and efficient solutions of these problems requires a well-designed combination of a suitable formulation/discretization, an effective solution algorithm (MLFMA version), and a carefully designed clustering mechanism.
Unknown
Exploring the importance and performance priorities of older adults with a user-centred approach to create a fall-free bathroom
(Wiley-Blackwell Publishing Ltd., 2024-06-19) Afacan, Yasemin; Barshan, Billur
Background Fall hazards in bathroom spaces constitute one of the most critical issues in the daily lives of older adults. Bathroom falls are somewhat different and constrained in nature than those in other parts of a home environment. Objectives This study aimed to adopt a user-centred approach to explore older adults' general bathroom needs, with a specific focus on showers and bathtubs as the designated activity area. Methods The authors employed an extended importance–performance analysis (IPA) with a mixed-method research design. Three hundred and eleven older adults participated in a face-to-face IPA questionnaire for the quantitative phase of the study. The authors gathered the qualitative data through open-ended questions from 59 older adults. Results The authors found positive correlation between older adults' attitudes towards an older-friendly bathroom and the potential for their bathrooms to be fall-free. The IPA calculations identify three key items with higher ratings in both importance and performance: The presence of appropriate artificial lighting, efficient mechanical ventilation and an accessible inside towel rail. Thematic analysis yields four themes: comfort, ease of access, error-proof design and emergency management. Conclusions The IPA calculations and thematic analysis confirm that older adults' rankings of importance and performance and their corresponding priority levels within the overarching themes indicate the need for these aspects to perform well and justify ongoing investments. The study concludes that addressing fall prevention requires not only designing specific solutions but also utilising appropriate technology in bathing and toileting activities. Implications for Practice Practitioners in geriatric and gerontological nursing, design, architecture and health care can use the importance and performance priority levels of older adults to guide the development and implementation of fall-free bathroom design. Policymakers can leverage the insights from this research to inform guidelines and regulations related to building codes, accessibility standards and healthcare policies.
Unknown
Exploiting residual errors in nonlinear online prediction
(Springer New York LLC, 2024-05-29) İlhan, Emirhan; Koç, Ahmet Berker; Kozat, Süleyman Serdar
We introduce a novel online (or sequential) nonlinear prediction approach that incorporates the residuals, i.e., prediction errors in the past observations, as additional features for the current data. Including the past error terms in an online prediction algorithm naturally improves prediction performance significantly since this information is essential for an algorithm to adjust itself based on its past errors. These terms are well exploited in many linear statistical models such as ARMA, SES, and Holts-Winters models. However, the past error terms are rarely or in a certain sense not optimally exploited in nonlinear prediction models since training them requires complex nonlinear state-space modeling. To this end, for the first time in the literature, we introduce a nonlinear prediction framework that utilizes not only the current features but also the past error terms as additional features, thereby exploiting the residual state information in the error terms, i.e., the model’s performance on the past samples. Since the new feature vectors contain error terms that change with every update, our algorithm jointly optimizes the model parameters and the feature vectors simultaneously. We achieve this by introducing new update equations that handle the effects resulting from the changes in the feature vectors in an online manner. We use soft decision trees and neural networks as the nonlinear prediction algorithms since these are the most widely used methods in highly publicized competitions. However, as we show, our methods are generic and any algorithm supporting gradient calculations can be straightforwardly used. We show through our experiments on the well-known real-life competition datasets that our method significantly outperforms the state-of-the-art. We also provide the implementation of our approach including the source code to facilitate reproducibility (https://github.com/ahmetberkerkoc/SDT-ARMA).
Unknown
Hybrid status update systems with dedicated and shared servers
(Institute of Electrical and Electronics Engineers Inc., 2024-10-24) Liyanaarachchi, Sahan; Ulukus, Sennur; Akar, Nail
Use of multi-path network topologies has become a prominent technique to assert timeliness in terms of age of information (AoI) and to improve resilience to link disruptions in communication systems. However, establishing multiple dedicated communication links among network nodes is a costly endeavor. Therefore, quite often, these secondary communication links are shared among multiple entities. Moreover, these multi-path networks come with the added challenge of out-of-order transmissions. In this paper, we study an amalgamation of the above two aspects, i.e., multi-path transmissions and link sharing. In this setting, we devise age-minimal scheduling schemes for a status update system composed of multiple sources sharing a single-server while having a separate dedicated server for each source so as to improve its timeliness. © 2024 International Federation for Information Processing - IFIP.
Open Access
Preface of the special issue: control, teams, and games
(Natural Sciences Publishing Corporation, 2024) Özbay, Hitay; Srikant, R.; Yuskel, S.
Open Access
Weakly confined organic-inorganic halide perovskite quantum dots as high-purity room-temperature single photon sources
(American Chemical Society, 2024-04-10) Wang, Bo; Lim, Jia Wei Melvin; Loh, Siow Mean; Mayengbam, Rishikanta; Ye, Senyun; Feng, Minjun; He, Huajun; Liang, Xiao; Cai, Rui; Zhang, Qiannan; Kwek, Leong-Chuan; Demir, Hilmi Volkan; Mhaisalkar, Subodh G.; Blundell, Steven A.; Chien Sum, Tze
Colloidal perovskite quantum dots (PQDs) have emerged as highly promising single photon emitters for quantum information applications. Presently, most strategies have focused on leveraging quantum confinement to increase the nonradiative Auger recombination (AR) rate to enhance single-photon (SP) purity in all-inorganic CsPbBr3 QDs. However, this also increases the fluorescence intermittency. Achieving high SP purity and blinking mitigation simultaneously remains a significant challenge. Here, we transcend this limitation with room-temperature synthesized weakly confined hybrid organic-inorganic perovskite (HOIP) QDs. Superior single photon purity with a low g((2))(0) < 0.07 +/- 0.03 and a nearly blinking-free behavior (ON-state fraction >95%) in 11 nm FAPbBr(3) QDs are achieved at room temperature, attributed to their long exciton lifetimes (tau(X)) and short biexciton lifetimes (tau(XX)). The significance of the organic A-cation is further validated using the mixed-cation FA(x)Cs(1-x)PbBr(3). Theoretical calculations utilizing a combination of the Bethe-Salpeter (BSE) and kp approaches point toward the modulation of the dielectric constants by the organic cations. Importantly, our findings provide valuable insights into an additional lever for engineering facile-synthesized room-temperature PQD single photon sources.
Open Access
VISPool: enhancing transformer encoders with vector visibility graph neural networks
(Association for Computational Linguistics, 2024-08-16) Alikaşifoğlu, Tuna; Aras, Arda Can; Koç, Aykut
The emergence of transformers has revolutionized natural language processing (NLP), as evidenced in various NLP tasks. While graph neural networks (GNNs) show recent promise in NLP, they are not standalone replacements for transformers. Rather, recent research explores combining transformers and GNNs. Existing GNN-based approaches rely on static graph construction methods requiring excessive text processing, and most of them are not scalable with the increasing document and word counts. We address these limitations by proposing a novel dynamic graph construction method for text documents based on vector visibility graphs (VVGs) generated from transformer output. Then, we introduce visibility pooler (VISPool), a scalable model architecture that seamlessly integrates VVG convolutional networks into transformer pipelines. We evaluate the proposed model on the General Language Understanding Evaluation (GLUE) benchmark datasets. VISPool outperforms the baselines with less trainable parameters, demonstrating the viability of the visibility-based graph construction method for enhancing transformers with GNNs. © 2024 Association for Computational Linguistics.
Open Access
(Smooth) Fictitious-play in identical-interest stochastic games with independent continuation-payoff estimates
(Natural Sciences Publishing Corporation, 2024) Zhang, K.Q.; Sayın, Muhammed Ömer; Ozdaglar, A.
In this paper, we study fictitious-play-type dynamics for identical-interest stochastic games (SGs) and show their convergence to the Nash equilibrium. We develop off-policy and on-policy dynamics, and generalize these learning dynamics and convergence results to the smooth fictitious play variant when the smooth best-response is used in the updates. One key feature of our dynamics is the independent estimates of the continuation payoffs among agents. While this feature makes the dynamics more natural and uncoupled, it also leads to the challenge that the auxiliary stage games encountered during learning can become non-identical-interest anymore. We handle such a deviation from the identical-interest setting by either focusing on specific structures, e.g., the single-controller or symmetric SGs, or studying specific sublinear stepsizes to characterize the convergence rate of such a deviation as timestep evolves.
Embargo
Demystifying trion emission in cdse nanoplatelets
(American Chemical Society, 2024-08-19) Riesner, Maurizio; Shabani, Farzan; Van Emmichoven, Levin Zeylmans; Klein, Julian; Delikanlı, Savaş; Fainblat, Rachel; Demir, Hilmi Volkan; Bacher, Gerd
At cryogenic temperatures, the photoluminescence spectrum of CdSe nanoplatelets (NPLs) usually consists of multiple emission lines, the origin of which is still under debate. While there seems to be consensus that both neutral excitons and trions contribute to the NPL emission, the prominent role of trions is rather puzzling. In this work, we demonstrate that Förster resonant energy transfer in stacks of NPLs combined with hole trap states in specific NPLs within the stack trigger trion formation, while single NPL spectra are dominated by neutral excitonic emission. This interpretation is verified by implementing copper (Cu+) dopants into the lattice as intentional hole traps. Trion emission gets strongly enhanced, and due to the large amount of hole trapping Cu+ states in each single NPL, trion formation does not necessarily require stacking of NPLs. Thus, the ratio between trion and neutral exciton emission can be controlled by either changing the amount of stacked NPLs during sample preparation or implementing copper dopants into the lattice which act as additional hole traps.
Open Access
Corrigendum to “natural language processing for defining linguistic features in schizophrenia: a sample from Turkish speakers” [Schizophr. Res. 266 (2024) 183–189]
(Elsevier BV, 2024-12) Çabuk, Tuğçe; Sevim, Nurullah; Mutlu, Emre; Yağcıoğlu, A. Elif Anıl; Koç, Aykut; Toulopoulou, Timothea
Open Access
Strongly stabilizing controller design for systems with time delay
(Natural Sciences Publishing Corporation, 2024) Özbay, Hitay
In this paper, some specific stable controller design techniques are reviewed for linear time invariant finite dimensional plants, and their extensions are discussed for systems with time delays. It is shown that under certain mild conditions, for strictly proper retarded delay systems with finitely many poles and zeros in C+, it is possible to obtain finite dimensional strongly stabilizing controllers. Illustrative examples are also given.
Open Access
Robustness of GaN on SiC low-noise amplifiers in common source and cascode configurations for X-band applications
(John Wiley & Sons Ltd., 2024-08) Nawaz, Muhammad Imran; Zafar, Salahuddin; Gürdal, Armağan; Akoğlu, Büşra Çankaya
Cascode HEMTs exhibit high gain and broadband performance. Promising reverse transmission makes matching networks simpler and insensitive to impedance on either side of the HEMT. On the other hand, common source (CS) HEMTs with intentional small inductance at the source provide simultaneous match for optimum noise and input impedance. This paper provides a performance comparison of 4 x 50 mu m cascode HEMTs-based low -noise amplifier and 4 x 50 mu m CS HEMTs-based low -noise amplifiers with specific emphasis on robustness, including survivability and reverse recovery time (RRT). Cascode LNA survives an input power of 33 dBm while CS LNA handles 30 dBm power, each having a 1 k Omega passive limiting resistor on the gate bias line. RRT of cascode LNA is also better. Better survivability and RRT for cascode LNA are attributed to its HEMT's stacked configuration. The designs of LNAs are described, along with their small -signal, noise, and large -signal characteristics in the X -band. Cascode LNA has a better input match, while CS LNA has a better output match. Gains are comparable, while CS LNA has better P1dB at higher band edge frequency. The noise figure for both LNAs is less than 1.9 dB, with CS LNA having a slight edge over cascode. This study benefits RF designers in choosing appropriate HEMT topology as per application for designing robust low -noise amplifiers.
Embargo
Hidden Semi-Markov Models for semantic-graph language modeling
(Elsevier Ltd, 2024-11) Yetim, Sadık Yağız; Duman, Tolga Mete; Arıkan, Orhan
Semantic communication is expected to play a critical role in reducing traffic load in future intelligent large-scale sensor networks. With advances in Machine Learning (ML) and Deep Learning (DL) techniques, design of semantically-aware systems has become feasible in recent years. This work focuses on improving the reliability of the semantic information represented in a graph-based language that has been recently proposed. Inaccuracies in the representation of the semantic information can arise due to multiple factors, such as algorithmic shortcomings or sensory errors, decreasing the performance of the semantic extractor. This study aims to model the temporal evolution of semantic information, represented using the graph language, to enhance its reliability. Each unique graph configuration is treated as a distinct state, leading to a Hidden Semi-Markov Model (HSMM) defined over the state space of the graph configurations. The HSMM formulation enables the integration of prior knowledge on the semantic signal into the graph sequences, enhancing the accuracy in identifying semantic innovations. Within the HSMM framework, algorithms designed for graph smoothing, semantic information fusion, and model learning are introduced. The efficacy of these algorithms in improving the reliability of the extracted semantic-graphs is demonstrated through simulations and video streams generated in the CARLA simulation environment.
Open Access
RIS-aided localization under pixel failures
(IEEE, 2024-08) Ozturk, Cuneyd; Keskin, Musa Furkan; Sciancalepore, Vincenzo; Wymeersch, Henk; Gezici, Sinan
Reconfigurable intelligent surfaces (RISs) hold great potential as one of the key technological enablers for beyond-5G wireless networks, improving localization and communication performance under line-of-sight (LoS) blockage conditions. However, hardware imperfections might cause RIS elements to become faulty, a problem referred to as pixel failures, which can constitute a major showstopper especially for localization. In this paper, we investigate the problem of RIS-aided localization of a user equipment (UE) under LoS blockage in the presence of RIS pixel failures, considering the challenging single-input single-output (SISO) scenario. We first explore the impact of such failures on accuracy through misspecified Cramér-Rao bound (MCRB) analysis, which reveals severe performance loss with even a small percentage of pixel failures. To remedy this issue, we develop two strategies for joint localization and failure diagnosis (JLFD) to detect failing pixels while simultaneously locating the UE with high accuracy. The first strategy relies on $\ell _{1}$-regularization through exploitation of failure sparsity. The second strategy detects the failures one-by-one by solving a multiple hypothesis testing problem at each iteration, successively enhancing localization and diagnosis accuracy. Simulation results show significant performance improvements of the proposed JLFD algorithms over the conventional failure-agnostic benchmark, enabling successful recovery of failure-induced performance degradations.