3586 research outputs found
Sort by
Attack robust aggregation in federated learning
No full textSon yıllarda, veri gizliliği ve güvenliği konusundaki artan endişeler makine öğrenimi modellerini merkezi bir veri havuzuna ihtiyaç duymadan eğitmeye olanak tanıyan federe öğrenme yaklaşımını ön plana çıkarmıştır. Ancak bu yenilikçi yaklaşım, dağıtık ve heterojen veri ortamlarında çalışmaktan kaynaklanan yeni güvenlik açıklarını da beraberinde getirmiştir. Özellikle sistemin güvenilirliğini ve performansını hedef alan düşmanca ataklar, federe öğrenme sistemlerinin dayanıklılığını tehdit etmektedir. Bu problemi çözmek amacıyla bu tezde düşmanca ataklara karşı Adaptive veRtex Mixup Federated Adversarial Training (ARM-FAT) adlı bir savunma mekanizması önerilmiştir. Önerilen yöntem temelinde bulundurduğu AVmixup algoritmasından elde edilen kaybı uyarlamalı olarak kontrol eden bir kat sayı, sağlamlığı temsil eden bir kayıp değeri ve temiz verileri temsil eden bir kayıp değeri sunmaktadır. Önerilen yöntem literatürde bulunan düşmanca eğitim yöntemleri ile karşılaştırılmıştır. Yöntemlerin karşılaştırılması farklı veri setleri, farklı düşmanca ataklar, farklı parametrelerde derin öğrenme mimarileri, farklı birleştirme algoritmaları ve farklı istemci sayıları üzerinden yapılmıştır. Elde edilen sonuçlar, ARM-FAT yönteminin birçok ayarda düşmanca ataklara karşı mevcut yöntemlerin başarımının üzerine çıkarak bir iyileştirme sağladığını göstermektedir. Bu yöntem daha da geliştirilerek gerçek hayat senaryosunda birçok federe öğrenme sisteminin güvenliğini sağlayan bir savunma mekanizması olarak kullanılabilir.In recent years, growing concerns about data privacy and security have brought the federated learning approach, which enables training machine learning models without the need for a centralized data pool, into prominence. However, this innovative approach has also introduced new security vulnerabilities arising from operating in distributed and heterogeneous data environments. In particular, adversarial attacks targeting the system's reliability and performance threaten the resilience of federated learning systems. To address this problem, this thesis proposes a defence mechanism called Adaptive veRtex Mixup Federated Adversarial Training (ARM-FAT) against adversarial attacks. The proposed method offers, based on its underlying AVmixup algorithm, a coefficient that adaptively controls the loss obtained from the algorithm, a loss value representing robustness, and a loss value representing clean data. The proposed method has been compared with adversarial training methods found in the literature. The comparison of the methods has been conducted across different datasets, different adversarial attacks, deep learning architectures with various parameters, different aggregation algorithms, and different numbers of clients. The results obtained demonstrate that the ARM-FAT method achieves an improvement by outperforming the performance of existing methods against adversarial attacks in many settings. This method can be further developed and used as a defence mechanism that ensures the security of many federated learning systems in real-life scenarios
Deception Detection in Turkish Hotel Reviews: A Comparative Study of Machine Learning and Deep Learning Approaches
No full textThis study presents a comparative analysis of machine learning and deep learning models for detecting deception in Turkish hotel reviews. For this purpose, fake, real and artificial intelligence-generated Turkish language hotel reviews are utilized. The real reviews dataset was created by filtering the data obtained from the Tripadvisor platform according to certain criteria, while two separate classes of deceptive reviews were established: one consisting of reviews written by human volunteers and another generated by artificial intelligence. The performances of machine learning and deep learning algorithms were tested for the detection of fake and AI-generated reviews. The results show that the BERTurk model achieved the highest performance with an F1-score of 0.93, followed by Artificial Neural Network (ANN), while Long Short Term Memory (LSTM), Support Vector Machine (SVM), and Random Forest (RF) also demonstrated strong classification capabilities. This study represents one of the first comprehensive deception detection studies in the Turkish language and contributes to the literature by demonstrating the effectiveness of transformer-based models for this task
Machine learning predictions of electro-optical properties in ZnO-doped nematic liquid crystals
No full textThis study explores the effect of zinc oxide (ZnO) nanomaterial doping on the electro-optical properties of 5CB-coded nematic liquid crystals and predicts these properties using machine learning algorithms. We produced seven composite structures with varying ZnO doping ratios and measured their electro-optical transmittance. Furthermore, a prediction model using four different machine learning algorithms (k-Nearest Neighbors, Decision Tree, Random Forest, and Extra Trees) was developed, which predicts optical transmittance as a function of voltage and doping ratio. The Extra Trees algorithm demonstrated the best prediction accuracy, achieving an R2 value of 91% on the experimental dataset. Subsequently, a new composite with a different doping ratio was then experimentally prepared and measured to validate the model, which was trained on the experimental dataset. This study highlights the utility of machine learning for predicting the electro-optical characteristics of doped liquid crystal structures, resulting in considerable time and resource savings in experimental procedures.Idot;zmir Bakimath;ray University Scientific Research Projects [TEZ.2021.004]; TUBITAK [BIDEB 2211-A]This study was supported by the project (no. TEZ.2021.004) accepted by Izmir Bakircay University BAP (Scientific Research Projects) Commission. Y Aygul thanks TUBITAK for their scholarship support under the BIDEB 2211-A programs
Search for dark matter produced in association with one or two top quarks in proton-proton collisions at = 13 TeV
No full textA search is performed for dark matter (DM) produced in association with a single top quark or a pair of top quarks using the data collected with the CMS detector at the LHC from proton-proton collisions at a center-of-mass energy of 13 TeV, corresponding to 138 fb?1 of integrated luminosity. An excess of events with a large imbalance of transverse momentum is searched for across 0, 1 and 2 lepton final states. Novel multivariate techniques are used to take advantage of the differences in kinematic properties between the two DM production mechanisms. No significant deviations with respect to the standard model predictions are observed. The results are interpreted considering a simplified model in which the mediator is either a scalar or pseudoscalar particle and couples to top quarks and to DM fermions. Axion-like particles that are coupled to top quarks and DM fermions are also considered. Expected exclusion limits of 410 and 380 GeV for scalar and pseudoscalar mediator masses, respectively, are set at the 95% confidence level. A DM particle mass of 1 GeV is assumed, with mediator couplings to fermions and DM particles set to unity. A small signal-like excess is observed in data, with the largest local significance observed to be 1.9 standard deviations for the 150 GeV pseudoscalar mediator hypothesis. Because of this excess, mediator masses are only excluded below 310 (320) GeV for the scalar (pseudoscalar) mediator. The results are also translated into model-independent 95% confidence level upper limits on the visible cross section of DM production in association with top quarks, ranging from 1 pb to 0.02 pb. © 2025 Elsevier B.V., All rights reserved
Parameter Predicting Postoperative Atrial Fibrillation in Coronary Artery Bypass Grafting Patients: TriglycerideCholesterol-Body Weight Index
No full textBackground: Postoperative atrial fibrillation (POAF) is a common complication after cardiac surgery, particularly coronary artery bypass grafting (CABG). Despite advances in surgical techniques, POAF remains a significant cause of morbidity and mortality. Objectives: This study investigates the potential of the Triglyceride-Cholesterol-Body weight Index (TCBI) as a predictor of POAF, focusing on the impact of nutritional status on surgical outcomes. Methods: This retrospective study included 321 patients who underwent CABG surgery between January 2010 and January 2024. TCBI was calculated using preoperative blood samples and compared between those who developed POAF and those who did not. Statistical analyses, including Cox regression and ROC analysis, were performed to assess the predictive value of TCBI for POAF. p<0.05 was considered statistically significant. Results: Patients who developed POAF had significantly lower TCBI (1790.8 +/- 689, 3413.3 +/- 1232, p<0.001, respectively) levels compared to those without POAF. Also, age (p<0.001), the frequency of hypertension (p=0.009), CRP (p=0.03), and WBC ( p=0.02) values were also significantly higher in patients who developed POAF.TCBI was identified as an independent predictor of POAF (OR: 0.998, 95% CI: 0.997-0.999, p<0.001)., with a cut-off value of 1932.4 predicting POAF with 75% sensitivity and 78% specificity. Conclusion: The TCBI is a reliable indicator for predicting POAF in CABG patients. Preoperative identification of patients with low TCBI could lead to targeted interventions, reducing postoperative complications and improving outcomes. Optimizing nutritional status before surgery may mitigate the risk of POAF
Uyku apnesi teşhisinde welch pediogramlarının derin öğrenme yöntemleri ile analizi
No full textElektrokardiyogram (EKG) sinyallerinden elde edilen Welch Periodogramlarının derin öğrenme ile analizine dayanan yeni bir uyku apnesi teşhis yöntemi bu çalışmada sunulmaktadır. Kullanılan tanı ölçütlerine göre yetişkin nüfusun %3 ila %17'sinde görülen uyku apnesi, günümüzde yetersiz teşhis edilmektedir. Bu durum, geleneksel tanı yöntemi olan polisomnografinin yüksek maliyeti, özel donanımlı laboratuvar gereksinimi ve hastanın doğal uyku düzenini etkileme potansiyeli gibi kısıtlamalardan kaynaklanmaktadır. Araştırmamızda geliştirilen otomatik tanı sistemi, Holter monitörlerinden alınan kayıtlardaki RR aralıklarının frekans spektrumundaki değişimleri Welch Periodogramları ile incelemektedir. Bu görsel temsiller, uyku apnesi esnasında kalp hızı değişkenliğinde meydana gelen karakteristik frekans alanı değişimlerini başarıyla yakalamaktadır. Önerilen metodoloji, Evrişimli Sinir Ağları (CNN) kullanarak bu spektral gösterimlerdeki uyku apnesine işaret eden desenleri tespit etmektedir. Sunduğumuz yaklaşımın invazif olmaması, ev ortamında uygulanabilirliği ve hasta konforunu artırması gibi önemli üstünlükleri bulunmaktadır. Çalışmamızın sonuçları, uyku tıbbı alanındaki metodolojik ilerlemelere katkı sağlarken, aynı zamanda uyku apnesi tanısında kolay erişilebilir araçların geliştirilmesine imkan tanımakta, böylece erken teşhis oranlarını ve hasta prognozunu iyileştirme potansiyeli taşımaktadır.A novel approach for diagnosing sleep apnea based on deep learning analysis of Welch Periodograms derived from electrocardiogram (ECG) signals is presented in this study. Sleep apnea, which affects between 3% and 17% of the adult population depending on diagnostic criteria used, is currently underdiagnosed. This situation stems from limitations of traditional diagnostic methods such as polysomnography, including high cost, requirement for specially equipped laboratories, and potential disruption of the patient's natural sleep patterns. The automated diagnostic system developed in our research examines frequency spectrum variations in RR intervals from Holter monitor recordings using Welch Periodograms. These visual representations successfully capture the characteristic frequency-domain changes in heart rate variability that occur during sleep apnea. The proposed methodology detects patterns indicative of sleep apnea in these spectral representations using Convolutional Neural Networks (CNN). Our approach offers significant advantages including being non-invasive, applicable in home environments, and enhancing patient comfort. The results of our study contribute to methodological advancements in sleep medicine while enabling the development of easily accessible tools for sleep apnea diagnosis, thus carrying potential to improve early detection rates and patient prognosis
Nanomechanical Properties and Application Potential of PbWO4 Crystals: Mechanical Behavior and Characterization
No full textThis study investigated the mechanical properties of a single crystal of lead tungstate (PbWO4) through nanoindentation measurements. The Oliver-Pharr method was used to determine the force-dependent Young's modulus and hardness of PbWO4. As the applied force increased, the values of hardness and Young's modulus decreased, which was attributed to the phenomenon known as the indentation size effect (ISE). The force-dependent data was analyzed using the proportional specimen resistance model, leading to a true hardness value of 2.87 GPa. By increasing the applied force from 5 to 100 mN, the Young's modulus decreased from 82.0 to 71.1 GPa. The observed reduction in Young's modulus is attributed to the formation of cracks within the material, which likely compromise its elastic response. The study also reported the contributions of plastic and elastic deformation components, revealing that plastic deformation was the dominant one. These results indicate that mechanical properties of PbWO4 make it a versatile material for technological applications in optoelectronics, photonics, radiation detection and scientific research.Scientific and Technological Research Council of Turkiye (TUBITAK)Open access funding provided by the Scientific and Technological Research Council of Turkiye (TUBITAK)
New Identities and Equation Solutions Involving k-Oresme and k-Oresme-Lucas Sequences
No full textNumber sequences are among the research areas of interest in both number theory and linear algebra. In particular, the study of matrix representations of recursive sequences is important in revealing the structural properties of these sequences. In this study, the relationships between the elements of the k-Fibonacci and k-Oresme sequences were analyzed using matrix algebra through matrix structures created by connecting the characteristic equations and roots of these sequences. In this context, using the properties of these matrices, the identities An2-An+1An-1=k-2n, An2-AnAn-1+1k2An-12=k-2n, and Bn2-BnBn-1+1k2Bn-12=-(k2-4)k-2n, and some generalizations such as Bn+m2-(k2-4)An-tBn+mAt+m-(k2-4)k2t-2nAt+m2=k-2m-2tBn-t2, At+m2-Bt-nAn+mAt+m+k2n-2tAn+m2=k-2n-2mAt-n2, and more were derived, where m,n,t is an element of & Zopf; and t not equal n. In addition to this, the solution pairs of the algebraic equations x2-Bpxy+k-2py2=k-2qAp2, x2-(k2-4)Apxy-(k2-4)k-2py2=k-2qBp2, and x2-Bpxy+k-2py2=-(k2-4)k-2qAp2 are presented, where Ap and Bp are k-Oresme and k-Oresme-Lucas numbers, respectively
Identification of low-momentum muons in the CMS detector using multivariate techniques in proton-proton collisions at ?s=13.6 TeV
No full textSoft muons with a transverse momentum below 10 GeV are featured in many processes studied by the CMS experiment, such as decays of heavy-flavor hadrons or rare tau lepton decays. Maximizing the selection efficiency for these muons, while simultaneously suppressing backgrounds from long-lived light-flavor hadron decays, is therefore important for the success of the CMS physics program. Multivariate techniques have been shown to deliver better muon identification performance than traditional selection techniques. To take full advantage of the large data set currently being collected during Run 3 of the CERN LHC, a new multivariate classifier based on a gradient-boosted decision tree has been developed. It offers a significantly improved separation of signal and background muons compared to a similar classifier used for the analysis of the Run 2 data. The performance of the new classifier is evaluated on a data set collected with the CMS detector in 2022 and 2023, corresponding to an integrated luminosity of 62 fb(-1).FWF; FNRS; FWO (Belgium); CNPq; CAPES; FAPERJ; FAPERGS; FAPESP (Brazil); BNSF (Bulgaria); MoST; NSFC (China); CSF (Croatia); RIF (Cyprus); SENESCYT (Ecuador); ERC PRG [MoER TK202]; Academy of Finland; MEC; CEA; CNRS/IN2P3 (France); SRNSF; BMBF; DFG; HGF (Germany); NKFIH (Hungary); DAE; DST; IPM; SFI (Ireland); INFN (Italy); NRF (Republic of Korea); MES (Latvia); MOE; UM (Malaysia); BUAP; CONACYT; UASLP-FAI (Mexico); PAEC (Pakistan); FCT (Portugal); MESTD (Serbia); PCTI (Spain); MOSTR (Sri Lanka); Swiss Funding Agencies (Switzerland); NSTDA; TUBITAK; DOE; NSF; Marie-Curie program; European Research Council; Horizon 2020 Grant [675440, 724704, 752730, 758316, 765710, 824093, 101115353, 101002207]; COST Action [CA16108]; Leventis Foundation; Alfred P. Sloan Foundation; Alexander von Humboldt Foundation; Science Committee [22rl-037]; Belgian Federal Science Policy Office; Fonds pour la Formation a la Recherche dans l'Industrie et dans l'Agriculture (FRIA-Belgium); FWO (Belgium) under the Excellence of Science -EOS [30820817]; Be.ing Municipal Science & Technology Commission [Z191100007219010]; Fundamental Research Funds for the Central Universities (China); Ministry of Education, Youth and Sports (MEYS) of the Czech Republic; Shota Rustaveli National Science Foundation [FR-22-985]; Deutsche Forschungsgemeinschaft (DFG) [EXC 2121, 400140256 -GRK2497]; Hellenic Foundation for Research and Innovation (HFRI) [2288]; Hungarian Academy of Sciences [K 131991, K 133046, K 138136, K 143460, K 143477, K 146913, K 146914, K 147048, 2020-2.2.1-ED-2021-00181, TKP2021-NKTA-64, 2021-4.1.2-NEMZ_KI-2024-00036]; Council of Science and Industrial Research, India - NextGenerationEU program (Italy); Latvian Council of Science; Ministry of Education and Science [2022/WK/14]; National Science Center [Opus 2021/41/B/ST2/01369, 2021/43/B/ST2/01552]; Fundacao para a Ciencia e a Tecnologia [CEECIND/01334/2018]; National Priorities Research Program by Qatar National Research Fund [MCIN/AEI/10.13039/501100011033]; ERDF a way of making Europe [MDM-2017-0765]; Programa Severo Ochoa del Principado de Asturias (Spain); National Science, Research and Innovation Fund via the Program Management Unit for Human Resources & Institutional Development, Research and Innovation [B39G670016]; Kavli Foundation; Nvidia Corporation; SuperMicro Corporation; Welch Foundation [C-1845]; Weston Havens Foundation (U.S.A.)We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC and thank the technical and administrative staffs at CERN and at other CMS institutes for their contributions to the success of the CMS effort. In addition, we gratefully acknowledge the computing centers and personnel of the Worldwide LHC Computing Grid and other centers for delivering so effectively the computing infrastructure essential to our analyses. Finally, we acknowledge the enduring support for the construction and operation of the LHC, the CMS detector, and the supporting computing infrastructure provided by the following funding agencies: SC (Armenia), BMBWF and FWF (Austria); FNRS and FWO (Belgium); CNPq, CAPES, FAPERJ, FAPERGS, and FAPESP (Brazil); MES and BNSF (Bulgaria); CERN; CAS, MoST, and NSFC (China); MINCIENCIAS (Colombia); MSES and CSF (Croatia); RIF (Cyprus); SENESCYT (Ecuador); ERC PRG, RVTT3 and MoER TK202 (Estonia); Academy of Finland, MEC, and HIP (Finland); CEA and CNRS/IN2P3 (France); SRNSF (Georgia); BMBF, DFG, and HGF (Germany); GSRI (Greece); NKFIH (Hungary); DAE and DST (India); IPM (Iran); SFI (Ireland); INFN (Italy); MSIP and NRF (Republic of Korea); MES (Latvia); LMTLT (Lithuania); MOE and UM (Malaysia); BUAP, CINVESTAV, CONACYT, LNS, SEP, and UASLP-FAI (Mexico); MOS (Montenegro); MBIE (New Zealand); PAEC (Pakistan); MES and NSC (Poland); FCT (Portugal); MESTD (Serbia); MCIN/AEI and PCTI (Spain); MOSTR (Sri Lanka); Swiss Funding Agencies (Switzerland); MST (Taipei); MHESI and NSTDA (Thailand); TUBITAK and TENMAK (Turkey); NASU (Ukraine); STFC (United Kingdom); DOE and NSF (U.S.A.). Individuals have received support from the Marie-Curie program and the European Research Council and Horizon 2020 Grant, contract Nos. 675440, 724704, 752730, 758316, 765710, 824093, 101115353, 101002207, and COST Action CA16108 (European Union); the Leventis Foundation; the Alfred P. Sloan Foundation; the Alexander von Humboldt Foundation; the Science Committee, project no. 22rl-037 (Armenia); the Belgian Federal Science Policy Office; the Fonds pour la Formation a la Recherche dans l'Industrie et dans l'Agriculture (FRIA-Belgium); the F.R.S.-FNRS and FWO (Belgium) under the Excellence of Science -EOS -be.h project n. 30820817; the Be.ing Municipal Science & Technology Commission, No. Z191100007219010 and Fundamental Research Funds for the Central Universities (China); the Ministry of Education, Youth and Sports (MEYS) of the Czech Republic; the Shota Rustaveli National Science Foundation, grant FR-22-985 (Georgia); the Deutsche Forschungsgemeinschaft (DFG), among others, under Germany's Excellence Strategy -EXC 2121 Quantum Universe -390833306, and under project number 400140256 -GRK2497; the Hellenic Foundation for Research and Innovation (HFRI), Project Number 2288 (Greece); the Hungarian Academy of Sciences, the New National Excellence Program -UNKP, the NKFIH research grants K 131991, K 133046, K 138136, K 143460, K 143477, K 146913, K 146914, K 147048, 2020-2.2.1-ED-2021-00181, TKP2021-NKTA-64, and 2021-4.1.2-NEMZ_KI-2024-00036 (Hungary); the Council of Science and Industrial Research, India; ICSC -National Research Center for High Performance Computing, Big Data and Quantum Computing and FAIR -Future Artificial Intelligence Research, funded by the NextGenerationEU program (Italy); the Latvian Council of Science; the Ministry of Education and Science, project no. 2022/WK/14, and the National Science Center, contracts Opus 2021/41/B/ST2/01369 and 2021/43/B/ST2/01552 (Poland); the Fundacao para a Ciencia e a Tecnologia, grant CEECIND/01334/2018 (Portugal); the National Priorities Research Program by Qatar National Research Fund; MCIN/AEI/10.13039/501100011033, ERDF a way of making Europe, and the Programa Estatal de Fomento de la Investigacion Cientifica y Tecnica de Excelencia Maria de Maeztu, grant MDM-2017-0765 and Programa Severo Ochoa del Principado de Asturias (Spain); the Chulalongkorn Academic into Its 2nd Century Project Advancement Project, and the National Science, Research and Innovation Fund via the Program Management Unit for Human Resources & Institutional Development, Research and Innovation, grant B39G670016 (Thailand); the Kavli Foundation; the Nvidia Corporation; the SuperMicro Corporation; the Welch Foundation, contract C-1845; and the Weston Havens Foundation (U.S.A.)
Exercise in Breast Cancer: A Key Factor for the Management of Side Effects of Breast Cancer
No full textExercise emerges as the most effective non-pharmacological intervention that is safe and effective in mitigating the side effects and functional impairments associated with cancer and multimodal treatment. The crucial effect of exercise on breast cancer patients is its positive influence on the rates of recurrence and mortality. Hence, individualized exercise prescriptions that are sustainable, patient-specific, and suitable for their needs are essential to improving clinical outcomes among breast cancer patients. Moderate-intensity aerobic exercise, resistance exercise, or combined exercise programs have been shown to improve the quality of life and alleviate diverse symptoms among breast cancer patients. Moreover, high-intensity interval training, mind body exercises, and aquatic exercise approaches have produced promising results on clinical outcomes in breast cancer patients. Nevertheless, further research is necessary to ascertain the safety, effectiveness, and feasibility of high-intensity interval training in cancer patients. Determining optimal exercise prescriptions is crucial, particularly emphasizing exercise frequency, intensity, type, and timing. A comprehensive assessment, including an analysis of concurrent symptoms and functional difficulties, should be conducted to develop suitable and individualized exercise prescriptions. Notwithstanding the beneficial effects of exercise on symptom management among breast cancer patients, exercise adherence is notably poor. To enhance adherence to exercise, conducting a comprehensive assessment to identify possible barriers is of utmost importance, in conjunction with patient education. Further research is required to determine the optimal exercise regimens regarding frequency, intensity, duration, and type to alleviate various symptoms associated with breast cancer and its multimodal treatments. © 2025 Elsevier B.V., All rights reserved