Publications without Fulltext
Permanent URI for this collectionhttps://hdl.handle.net/20.500.14288/3
Browse
649 results
Filters
Advanced Search
Filter by
Settings
Search Results
Item Metadata only The use of class imbalanced learning methods on ULSAM data to predict the case-control status in genome-wide association studies(Springernature, 2023) 0000-0002-9847-4030; 0000-0001-6981-6962; Morris, Andrew P.; Tasdelen, Bahar; N/A; N/A; Öztornacı, Ragıp Onur; Syed, Hamzah; Researcher; Faculty Member; Koç University Research Center for Translational Medicine (KUTTAM) / Koç Üniversitesi Translasyonel Tıp Araştırma Merkezi (KUTTAM); N/A; School of Medicine; N/A; 318138Machine learning (ML) methods for uncovering single nucleotide polymorphisms (SNPs) in genome-wide association study (GWAS) data that can be used to predict disease outcomes are becoming increasingly used in genetic research. Two issues with the use of ML models are finding the correct method for dealing with imbalanced data and data training. This article compares three ML models to identify SNPs that predict type 2 diabetes (T2D) status using the Support vector machine SMOTE (SVM SMOTE), The Adaptive Synthetic Sampling Approach (ADASYN), Random under sampling (RUS) on GWAS data from elderly male participants (165 cases and 951 controls) from the Uppsala Longitudinal Study of Adult Men (ULSAM). It was also applied to SNPs selected by the SMOTE, SVM SMOTE, ADASYN, and RUS clumping method. The analysis was performed using three different ML models: (i) support vector machine (SVM), (ii) multilayer perceptron (MLP) and (iii) random forests (RF). The accuracy of the case-control classification was compared between these three methods. The best classification algorithm was a combination of MLP and SMOTE (97% accuracy). Both RF and SVM achieved good accuracy results of over 90%. Overall, methods used against unbalanced data, all three ML algorithms were found to improve prediction accuracy.Item Metadata only Making opportunity sales in attended home delivery(Pergamon-Elsevier Science Ltd, 2023) 0000-0002-3839-8371; 0000-0001-5076-9561; Arslan, Okan; Laporte, Gilbert; Department of Industrial Engineering; Yıldız, Barış; Ötken, Çelen Naz; Faculty Member; Master Student; College of Engineering; 258791Research on time window management in attended home delivery mainly focuses on influencing customers' delivery time choices to reduce monetary and environmental costs. In this study, we adopt a different perspective and propose exploiting otherwise idle time and vehicle capacities to generate extra profits through opportunity sales for e-groceries. We consider nudging potential target customers (residing in locations that are "easy" to insert into the delivery tours) with push notifications to generate new sales. These customers are incentivized for purchases by dropping the terms imposed on standard e-grocery sales such as service fees or minimum purchase quantities. Managing the delivery operations for this innovative business model requires concurrently choosing the target customers and planning the vehicle routes under the offer acceptance and response time uncertainties. To solve this challenging problem, we propose an integer linear programming model that enables a decomposition of the problem into a routing problem and several customer selection problems. For the solution to the customer selection problems, we propose mathematical models with varying risk-taking levels. We also investigate the benefits of dynamic policies to take advantage of the information revealed during the delivery operations in order to adjust customer selection and vehicle routing decisions. Our extensive numerical experiments show that these models, equipped with dynamic decision making, can compete with the risk-ignorant models for the total profit while generating more sales per offer, as well as ensuring timely execution of the delivery operations.Item Metadata only An overview: steady-state quantum entanglement via reservoir engineering(Old City Publishing Inc, 2023) 0000-0002-9134-3951; 0000-0001-7411-3399; Department of Physics; N/A; Müstecaplıoğlu, Özgür Esat; Pedram, Ali; Faculty Member; PhD Stuent; College of Sciences; Graduate School of Sciences and Engineering; 1674; N/AWe present a short overview of quantum entanglement generation and preservation in a steady state. In addition to the focus on quantum entanglement stabilization, we briefly discuss the same objective for steadystate quantum coherence. The overview classifies the approaches into two main categories: hybrid drive and dissipation methods and purely dissipative schemes. Furthermore, purely dissipative schemes are discussed under two subclasses of equilibrium and nonequilibrium environments. The significance of the dissipative route to sustained quantum entanglement and challenges against it are pointed out. Besides the value of steady-state entanglement for existing quantum technologies, quantum computation, communication, sensing, and simulation, its unique opportunities for emerging and future quantum technology applications, particularly quantum heat engines and quantum energy processing, are discussed.Item Metadata only A novel approach to quantify microsleep in drivers with obstructive sleep apnea by concurrent analysis of EEG patterns and driving attributes(IEEE-Inst Electrical Electronics Engineers Inc, 2024) 0000-0001-9067-6538; 0000-0002-7544-5974; 0000-0002-4041-4529; 0000-0001-5575-2195; 0000-0002-8286-7956; N/A; Department of Electrical and Electronics Engineering; N/A; N/A; N/A; Peker, Yüksel; Gürsoy, Beren Semiz; Çelik, Yeliz; Arbatlı, Semih; Minhas, Riaz; Faculty Member; Faculty Member; Researcher; PhD Student; Master Student; School of Medicine; College of Engineering; N/A; Graduate School of Health Sciences; Graduate School of Sciences and Engineering; 234103; 332403; N/A; N/A; N/AAccurate quantification of microsleep (MS) in drivers is crucial for preventing real-time accidents. We propose one-to-one correlation between events of high-fidelity driving simulator (DS) and corresponding brain patterns, unlike previous studies focusing general impact of MS on driving performance. Fifty professional drivers with obstructive sleep apnea (OSA) participated in a 50-minute driving simulation, wearing six-channel Electroencephalography (EEG) electrodes. 970 out-of-road OOR (microsleep) events (wheel and boundary contact >= 1 s), and 1020 on-road OR (wakefulness) events (wheel and boundary disconnection >= 1 s), were recorded. Power spectrum density, computed using discrete wavelet transform, analyzed power in different frequency bands and theta/alpha ratios were calculated for each event. We classified OOR (microsleep) events with higher theta/alpha ratio compared to neighboring OR (wakefulness) episodes as true MS and those with lower ratio as false MS. Comparative analysis, focusing on frontal brain, matched 791 of 970 OOR (microsleep) events with true MS episodes, outperforming other brain regions, and suggested that some unmatched instances were due to driving performance, not sleepiness. Combining frontal channels F3 and F4 yielded increased sensitivity in detecting MS, achieving 83.7% combined mean identification rate (CMIR), surpassing individual channel's MIR, highlighting potential for further improvement with additional frontal channels. We quantified MS duration, with 95% of total episodes lasting between 1 to 15 seconds, and pioneered a robust correlation (r = 0.8913, p<0.001) between maximum drowsiness level and MS density. Validating simulator's signals with EEG patterns by establishing a direct correlation improves reliability of MS identification for assessing fitness-to-drive of OSA-afflicted adults.Item Metadata only Investigation of lattice infill parameters for additively manufactured bone fracture plates to reduce stress shielding(Pergamon-Elsevier Science Ltd, 2023) 0000-0002-8316-9623; 000-0002-8383-6000; Karaismailoglu, Bedri; Ashkani-Esfahani, Soheil; Department of Mechanical Engineering; N/A; Lazoğlu, İsmail; Subaşı, Ömer; Faculty Member; PhD Student; Manufacturing and Automation Research Center (MARC); College of Engineering; Graduate School of Sciences and Engineering; 179391Background: Stress shielding is a detrimental phenomenon caused by the stiffness mismatch between metallic bone plates and bone tissue, which can hamper fracture healing. Additively manufactured plates can decrease plate stiffness and alleviate the stress shielding effect. Methods: Rectilinear lattice plates with varying cell sizes, wall thicknesses, and orientations are computationally generated. Finite element analysis is used to calculate the four-point bending stiffness and strength of the plates. The mechanical behaviors of three different lattice plates are also simulated under a simple diaphyseal fracture fixation scenario. Results: The study shows that with different combinations of lattice infill parameters, plates with up to 68% decrease in stiffness compared to the 100% infill plate can be created. Moreover, in the fixation simulations, the least stiff lattice plate displays 53% more average stress distribution at the healing callus region compared to the 100% infill plate. Conclusions: Using computational techniques, it has been demonstrated that additively manufactured stiffness-reduced bone plates can successfully address stress shielding with the strategic modulation of lattice infill pa-rameters. Lattice plates with design versatility have the potential for use in various fracture fixation scenarios.Item Metadata only Design and modeling of a PVDF-TrFe flexible wind energy harvester(Tubitak Scientific and Technological Research Council Turkey, 2023) 0000-0002-9777-6619; 0000-0002-0784-1537; Department of Mechanical Engineering; N/A; Beker, Levent; Kullukçu, Berkay; Faculty Member; Master Student; College of Engineering; Graduate School of Sciences and Engineering; 308798; N/AThis study presents the simulation, experimentation , design considerations of a Poly(vinylidene fluoride co-trifluoroethylene)/ Polyethylene Terephthalate (PVDF-TrFe / PET), laser-cut, flexible piezoelectric energy harvester. It is possible to obtain energy from the environment around autonomous sensor systems, which can then be used to power various equipment. This article investigates the actuation means of ambient vibration, which is a good candidate for using piezoelectric energy harvester (PEH) devices. The output voltage characteristics were analyzed in a wind test apparatus. Finite element modeling (FEM) was done for von Mises stress , modal analysis. Resonance frequency sweeps, quality factors, and damping ratios of the circular plate were given numerically. For a PVDF-TrFe piezoelectric layer thickness of 18 mu m and 1.5 mm radius, a damping ratio of 0.117 and a quality factor of 4.284 was calculated. Vmax was calculated as 984 mV from the wind setup experiments and compared with the FEM outputs.Item Metadata only Multi-field de-interlacing using deformable convolution residual blocks and self-attention(IEEE, 2022) 0000-0003-1465-8121; 0000-0001-6840-5766; Department of Electrical and Electronics Engineering; N/A; Tekalp, Ahmet Murat; Ji, Ronglei; Faculty Member; PhD Student; College of Engineering; Graduate School of Sciences and Engineering; 26207; N/AAlthough deep learning has made significant impact on image/video restoration and super-resolution, learned deinterlacing has so far received less attention in academia or industry. This is despite deinterlacing is well-suited for supervised learning from synthetic data since the degradation model is known and fixed. In this paper, we propose a novel multi-field full frame-rate deinterlacing network, which adapts the state-of-the-art superresolution approaches to the deinterlacing task. Our model aligns features from adjacent fields to a reference field (to be deinterlaced) using both deformable convolution residual blocks and self attention. Our extensive experimental results demonstrate that the proposed method provides state-of-the-art deinterlacing results in terms of both numerical and perceptual performance. At the time of writing, our model ranks first in the Full FrameRate LeaderBoard at https://videoprocessing.ai/benchmarks/deinterlacer.htmlItem Metadata only Multi-GPU communication schemes for iterative solvers: when CPUs are not in charge(Association for Computing Machinery, 2023) 0000-0002-2351-0770; 0000-0002-9603-2466; 0000-0001-7235-6418; N/A; Wahib, Mohamed; Department of Computer Engineering; N/A; N/A; N/A; Erten, Didem Unat; Sağbili Doğan; Baydamirli Javid; Ismayilov, Ismayil; Faculty Member; PhD Student; PhD Student; Master Student; College of Engineering; Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; 219274; N/A; N/A; N/AThis paper proposes a fully autonomous execution model for multi-GPU applications that completely excludes the involvement of the CPU beyond the initial kernel launch. In a typical multi-GPU application, the host serves as the orchestrator of execution by directly launching kernels, issuing communication calls, and acting as a synchronizer for devices. We argue that this orchestration, or control flow path, causes undue overhead and can be delegated entirely to devices to improve performance in applications that require communication among peers. For the proposed CPU-free execution model, we leverage existing techniques such as persistent kernels, thread block specialization, device-side barriers, and device-initiated communication routines to write fully autonomous multi-GPU code and achieve significantly reduced communication overheads. We demonstrate our proposed model on two broadly used iterative solvers, 2D/3D Jacobi stencil and Conjugate Gradient(CG). Compared to the CPU-controlled baselines, the CPU-free model can improve 3D stencil communication latency by 58.8% and provide a 1.63x speedup for CG on 8 NVIDIA A100 GPUs. The project code is available at https://github.com/ParCoreLab/CPU-Free-model. © 2023 Owner/Author(s).Item Metadata only Defending against targeted poisoning attacks in federated learning(IEEE Computer Soc, 2022) 0000-0002-7676-0167; N/A; Department of Computer Engineering; Department of Computer Engineering; Gürsoy, Mehmet Emre; Erbil Pınar; Faculty Member; Undergraduate Student; College of Engineering; College of Engineering; 330368; N/AFederated learning (FL) enables multiple participants to collaboratively train a deep neural network (DNN) model. To combat malicious participants in FL, Byzantine-resilient aggregation rules (AGRs) have been developed. However, although Byzantine-resilient AGRs are effective against untargeted attacks, they become suboptimal when attacks are stealthy and targeted. In this paper, we study the problem of defending against targeted data poisoning attacks in FL and make three main contributions. First, we propose a method for selective extraction of DNN parameters from FL participants' update vectors that are indicative of attack, and embedding them into low-dimensional latent space. We show that the effectiveness of Byzantine-resilient AGRs such as Trimmed Mean and Krum can be improved if they are used in combination with our proposed method. Second, we develop a clustering-based defense using X-Means for separating items into malicious versus benign clusters in latent space. Such separation allows identification of malicious versus benign updates. Third, using the separation from the previous step, we show that a "clean" model (i.e., a model that is not negatively impacted by the attack) can be trained using only the benign updates. We experimentally evaluate our defense methods on Fashion-MNIST and CIFAR-10 datasets. Results show that our methods can achieve up to 95% true positive rate and 99% accuracy in malicious update identification across various settings. In addition, the clean models trained using our approach achieve similar accuracy compared to a baseline scenario without poisoning.Item Metadata only Utility-aware and privacy-preserving mobile query services(IEEE Computer Soc, 2023) 0000-0002-7676-0167; Yigitoglu, Emre; Liu, Ling; Department of Computer Engineering; Gürsoy, Mehmet Emre; Faculty Member; College of Engineering; 330368Location-based queries enable fundamental services for mobile users. While the benefits of location-based services (LBS) are numerous, exposure of mobile users' locations to untrusted LBS providers may lead to privacy concerns. This article proposes StarCloak, a utility-aware and attack-resilient location anonymization service for privacy-preserving LBS usage. StarCloak combines several desirable properties. First, unlike conventional approaches which are indifferent to underlying road network structure, StarCloak uses the concept of stars and proposes cloaking graphs for effective location cloaking on road networks. Second, StarCloak supports user-specified $k$k-user anonymity and $l$l-segment indistinguishability, for enabling personalized privacy protection and for serving users with varying privacy preferences. Third, StarCloak achieves strong attack-resilience against replay and query injection attacks through randomized star selection and pruning. Finally, to enable efficient query processing with high throughput and low bandwidth overhead, StarCloak makes cost-aware star selection decisions by considering query evaluation and network communication costs. We evaluate StarCloak on two datasets using real-world road networks, under various privacy and utility constraints. Results show that StarCloak achieves improved query success rate and throughput, reduced anonymization time and network usage, and higher attack-resilience in comparison to XStar, its most relevant competitor.