5862 research outputs found
Sort by
Olumsuz Duygu Düzenleme Beklentileri Ölçe ̆gi Türkçe Formunun Psikometrik Özellikleri
No full text[No Abstract Available
Courier Payout Cash-Flow Prediction in Crowdsourced E-Commerce Logistics: a Hybrid Machine Learning Approach
No full textIn the rapidly growing sector of crowdsourced e-commerce logistics, where delivery volumes are highly variable, the effective management of courier payouts becomes essential to maintain operational efficiency. This paper introduces a comprehensive hybrid approach, blending clustering methods with multiple advanced regression models, to accurately predict daily courier payout cash-flows. By utilizing real-world data from e-commerce operations, our methodology estimates the daily financial outflows for courier payments, a critical component for adapting to the dynamic and unpredictable nature of crowdsourced logistics. Our approach includes a thorough comparative analysis of several stateof- the-art regression models - namely, XGBoost Regressor, LightGBM Regressor, and Facebook's PROPHET - in conjunction with clustering techniques that categorize similar cross-docks based on distinct characteristics. This integrated, hybrid strategy aims to provide precise daily financial predictions for each cross-dock, which is crucial for robust financial planning and effective resource allocation. The practical implications of this research are significant, offering logistics companies a powerful tool to navigate the complexities of e-commerce environments. By ensuring more accurate cash-flow predictions, companies can optimize their operations, reduce financial uncertainties, and improve overall service quality in the highly competitive and fast-paced world of e-commerce logistics.Türkiye Bilimsel ve Teknolojik Araştırma Kurumu, TÜBİTAKTUBITAK; HepsiJetSupported by TUBITAK and HepsiJet.Conference Proceedings Citation Index - Scienc
Multi–label Emotion Classification With Fine-Tuned Bert Andcontrastive Learning
No full textDuygular insan etkileşiminin temelini oluşturmakta ve bunların sınıflandırılması ve tanınması doğal dil işlemede (NLP). önemli zorluklar yaratmaktadır. Bu araştırma, metinsel verilerde çok etiketli duygu sınıflandırması için karşılaştırmalı öğrenme ile BERT tabanlı örneklenmemiş (uncased) modellerin ince ayarının etkinliğini değerlendirmektedir. SemEval 2024 Görev 3, Alt Görev 1 için sağlanan, Ekman'ın temel duygularını (Sevinç, Üzüntü, Öfke, Korku, İğrenme ve Şaşkınlık) içeren 1.374 elle işaretlemeli konuşmayı içeren durum komedisi Friends'ten alınan verileri kullanarak, ContrastiveBERT'in nüanslı duygusal durumları daha iyi yakalayıp yakalayamayacağını, sınıf dengesizliklerini ele alıp alamayacağını ve bu altı duyguyu sınıflandırmada standart BERT'ten daha iyi performans gösterip gösteremeyeceği değerlendirilmiştir. ContrastiveBERT yaklaşımı, temel BERT modeline kıyasla %9,1 daha yüksek F1 puanı, %11,3 ROC AUC artışı ve %10,17 doğruluk artışı ile önemli performans iyileştirmeleri göstermiştir. Bu araştırma, BERT'in tek bir metin parçası içerisinde birden fazla duyguyu yakalamadaki performansına ilişkin anlayışımızı geliştirmeye katkıda bulunarak, duygu analizinde daha geniş bir uygulama alanının önünü açmaktadırEmotions are fundamental to human interaction, and their classification and recognition pose significant challenges in natural language processing (NLP). This research evaluates the effectiveness of fine-tuning BERT-base uncased models with contrastive learning for multi-label emotion classification in textual data. Using data from the sitcom Friends, which includes 1,374 manually annotated conversations featuring Ekman's basic emotions—Joy, Sadness, Anger, Fear, Disgust, and Surprise—provided for SemEval 2024 Task 3, Subtask 1, we assess whether ContrastiveBERT can better capture nuanced emotional states, address class imbalances, and outperform standard BERT in classifying these six emotions. The ContrastiveBERT approach demonstrated notable performance improvements, with a 9.1% higher F1 score, 11.3% increase in ROC AUC, and a 10.17% improvement in accuracy compared to the baseline BERT model. This research contributes to enhancing our understanding of ContrastiveBERT's performance in capturing multiple emotions within a single text segment, paving the way for its broader application in emotion analysi
International Collaborations Among Schools as a Follow-Up of the International Year of Sound
No full textIn this paper, moving from the successful initiatives connected to the International Year of Sound and a solid international collaboration developed during the past ten years in the frame of INAD, the International Noise Awareness days, the authors present a new project that involves schools from Italy, Spain and Turkey together with the three National Acoustical Societies. The project regards education in acoustics at schools and co-design of silent solutions has been object of an application for grants under the call Erasmus KA220-SCH - cooperation partnerships in school education
Trauma, Posttraumatic Growth, and World Literature: Metamorphoses and a Literary Arts Praxis.
No full text[No Abstract Available]Emerging Sources Citation Inde
Probabilistic Approach To Assess and Minimize the Voltage Violation Risk in Active Distribution Networks
No full text100% Pure New Zealand; AUT; Centre for Future Power and Energy Research; et al.; IEEE New Zealand North Section; New Zealand TourismThe increasing trend in using renewable energy resources in distribution systems has encouraged system operators to find the best methods to decrease the growing uncertainty's impact on system operation. A probabilistic approach based on the combination of Monte Carlo simulation and Particle Swarm Algorithm is proposed in this paper to reduce the risk of voltage magnitude violations. Also, a novel criterion is used to assess the risk of voltage magnitude violations in distribution system operation. This index is based on providing voltage samples using a probabilistic approach. Therefore, enhancing the confidence level of voltage risk is considered an objective function in finding the optimum location of energy storage systems. The proposed approach is applied to the IEEE 33-bus test system, and the results show that two ESS units installed at appropriate locations can solve all the voltage magnitude violation problems.Scientific and Technological Research Council of Turkey, TUBITAK [1059B212200475]This research is funded as a part of 1059B212200475 Probabilistic assessment for a comprehensive design of Forthcoming Turkish Distribution Network Integrating Electric Vehicles, Energy Storage Systems, and Renewable Generation of 2221 Project organized by The Scientific and Technological Research Council of Turkey, TUBITAK.Conference Proceedings Citation Index - Scienc
Decoding Functional Brain Data for Emotion Recognition: A Machine Learning Approach
No full textBalli, Tugce/0000-0002-6509-3725; Tulay, Emine Elif/0000-0003-0150-5476The identification of emotions is an open research area and has a potential leading role in the improvement of socio-emotional skills such as empathy, sensitivity, and emotion recognition in humans. The current study aimed at using Event Related Potential (ERP) components (N100, N200, P200, P300, early Late Positive Potential (LPP), middle LPP, and late LPP) of EEG data for the classification of emotional states (positive, negative, neutral). EEG datawere collected from 62 healthy individuals over 18 electrodes. An emotional paradigm with pictures from the International Affective Picture System (IAPS) was used to record the EEG data. A linear Support Vector Machine (C = 0.1) was used to classify emotions, and a forward feature selection approach was used to eliminate irrelevant features. The early LPP component, which was the most discriminative among all ERP components, had the highest classification accuracy (70.16%) for identifying negative and neutral stimuli. The classification of negative versus neutral stimuli had the best accuracy (79.84%) when all ERP components were used as a combined feature set, followed by positive versus negative stimuli (75.00%) and positive versus neutral stimuli (68.55%). Overall, the combined ERP component feature sets outperformed single ERP component feature sets for all stimulus pairings in terms of accuracy. These findings are promising for further research and development of EEG-based emotion recognition systems.Science Citation Index Expande
Boosting biomolecular switch efficiency with quantum coherence
No full textThe resource theory of quantum thermodynamics has emerged as a powerful tool for exploring the outof -equilibrium dynamics of microscopic and highly correlated systems. Recently, it has been employed in photoisomerization, a mechanism facilitating vision through the isomerism of the photoreceptor protein rhodopsin, to elucidate the fundamental limits of efficiency inherent in this physical process. Limited attention has been given to the impact of energetic quantum coherences in this process, as these coherences do not influence the energy -level populations within an individual molecule subjected to thermal operations. However, a specific type of energetic quantum coherences can impact the energy -level populations in the scenario involving two or more molecules. In this study, we examine the case of two molecules undergoing photoisomerization to show that energetic quantum coherence can function as a resource that amplifies the efficiency of photoisomerization. These insights offer evidence for the role of energetic quantum coherence as a key resource in the realm of quantum thermodynamics at mesoscopic scales.Gordon and Betty Moore Foundation, Lillian Martin; Oxford Martin School; John Fell Fund; Scientific and Technological Research Council of Turkey (TUBITAK) [120F089]; ENS Paris-Saclay ARPE programmeWe thank Prof. M. Olivucci and Dr. L. Pedraza-Gonzalez for helpful discussions on their simulations of rhodopsin and for sharing simulated parameters. T.F. thanks the Gordon and Betty Moore Foundation, Lillian Martin and the Oxford Martin School, and the John Fell Fund for support. O.P. acknowledges support by the Scientific and Technological Research Council of Turkey (TUBITAK) under Grant No. 120F089. M.B. thanks the ENS Paris-Saclay ARPE programme for support.Science Citation Index Expande
Axial, planar-diagonal, body-diagonal fields on the cubic-spin spin glass in d=3: A plethora of ordered phases under finite fields
No full textA nematic phase, previously seen in the d = 3 classical Heisenberg spin-glass system, occurs in the n-component cubic-spin spin-glass system, between the low-temperature spin-glass phase and the hightemperature disordered phase, for number of spin components n >= 3, in spatial dimension d = 3, thus constituting a liquid-crystal phase in a dirty (quenched-disordered) magnet. Furthermore, under application of a variety of uniform magnetic fields, a veritable plethora of phases is found. Under uniform magnetic fields, 17 different phases and two spin-glass phase diagram topologies (meaning the occurrences and relative positions of the many phases), qualitatively different from the conventional spin-glass phase diagram topology, are seen. The chaotic rescaling behaviors and their Lyapunov exponents are calculated in each of these spin-glass phase diagram topologies. These results are obtained from renormalization-group calculations that are exact on the d = 3 hierarchical lattice and, equivalently, approximate on the cubic spatial lattice. Axial, planar-diagonal, or body-diagonal finite-strength uniform fields are applied to n = 2 and 3 component cubic-spin spin-glass systems in d=3.Kadir Has University Doctoral Studies Scholarship Fund; Academy of Sciences of Turkey (TUEBA)Support by the Kadir Has University Doctoral Studies Scholarship Fund and by the Academy of Sciences of Turkey (TUEBA) is gratefully acknowledged.Science Citation Index Expande