Organizational Unit: Computer Science
Loading...
Date established
City
Country
ID
490 results
Publication Search Results
Now showing 1 - 10 of 490
ArticlePublication Open Access The effect of appearance of virtual agents in human-agent negotiation(MDPI, 2022-09) Türkgeldi, Berkay; Özden, Cana Su; Aydoğan, Reyhan; Computer Science; AYDOĞAN, Reyhan; Türkgeldi, Berkay; Özden, Cana SuArtificial Intelligence (AI) changed our world in various ways. People start to interact with a variety of intelligent systems frequently. As the interaction between human and AI systems increases day by day, the factors influencing their communication have become more and more important, especially in the field of human-agent negotiation. In this study, our aim is to investigate the effect of knowing your negotiation partner (i.e., opponent) with limited knowledge, particularly the effect of familiarity with the opponent during human-agent negotiation so that we can design more effective negotiation systems. As far as we are aware, this is the first study investigating this research question in human-agent negotiation settings. Accordingly, we present a human-agent negotiation framework and conduct a user experiment in which participants negotiate with an avatar whose appearance and voice are a replica of a celebrity of their choice and with an avatar whose appearance and voice are not familiar. The results of the within-subject design experiment show that human participants tend to be more collaborative when their opponent is a celebrity avatar towards whom they have a positive feeling rather than a non-celebrity avatar.Conference paperPublication Open Access Evidential deep learning to quantify classification uncertainty(Neural Information Processing Systems Foundation, 2018) Şensoy, Murat; Kaplan, L.; Kandemir, M.; Computer Science; ŞENSOY, MuratDeterministic neural nets have been shown to learn effective predictors on a wide range of machine learning problems. However, as the standard approach is to train the network to minimize a prediction loss, the resultant model remains ignorant to its prediction confidence. Orthogonally to Bayesian neural nets that indirectly infer prediction uncertainty through weight uncertainties, we propose explicit modeling of the same using the theory of subjective logic. By placing a Dirichlet distribution on the class probabilities, we treat predictions of a neural net as subjective opinions and learn the function that collects the evidence leading to these opinions by a deterministic neural net from data. The resultant predictor for a multi-class classification problem is another Dirichlet distribution whose parameters are set by the continuous output of a neural net. We provide a preliminary analysis on how the peculiarities of our new loss function drive improved uncertainty estimation. We observe that our method achieves unprecedented success on detection of out-of-distribution queries and endurance against adversarial perturbations.
Conference paperPublication Metadata only Büyük veri problemlerine çözüm olarak veri akış madenciliği(IEEE, 2013) Ölmezoğulları, Erdi; Arı, İsmail; Çelebi, Ö. F.; Ergüt, S.; Computer Science; ARI, Ismail; Ölmezoğulları, ErdiGünümüzde bilişim dünyası faydalı bilgiye ulaşma yolunda “büyük veri” problemleri (verinin kütlesi, hızı, çeşitliliği, tutarsızlığı) ile baş etmeye çalışmaktadır. Bu makalede, büyük veri akışları üzerinde İlişkisel Kural Madenciliği’nin (İKM) daha önce literatürde yapılmamış bir şekilde “çevrimiçi” olarak gerçeklenme detayları ile başarım bulguları paylaşılacaktır. Akış madenciliği için Apriori ile FP-Growth algoritmaları Esper isimli olay akış motoruna eklenmiştir. Elde edilen sistem üzerinde bu iki algoritma kayan penceler ve LastFM sosyal müzik sitesi verileri kullanılarak karşılaştırılmıştır. Başarımı yüksek olan FPGrowth seçilerek gerçek-zamanlı ve kural-tabanlı bir tavsiye motoru oluşturulması sağlanmıştır. En önemli bulgularımız çevrimiçi kural çıkarımı sayesinde: (1) çevrimdışı kural çıkarımından çok daha fazla kuralın (2) çok daha hızlı ve etkin olarak ve (3) çok daha önceden hesaplanabileceği gösterilmiştir. Ayrıca müzik zevklerine uygun “George Harrison⇒The Beatles” gibi pekçok ilginç ve gerçekçi kural bulunmuştur. Sonuçlarımızın ileride diğer büyük veri analitik sistemlerinin tasarım ve gerçeklemesine ışık tutacağını ummaktayız.Conference paperPublication Metadata only COSMOS on steroids: a Cheap detector for cheapfakes(The ACM Digital Library, 2021) Akgül, T.; Civelek, Tuğçe Erkılıç; Uğur, Deniz; Beğen, Ali Cengiz; Computer Science; BEĞEN, Ali Cengiz; Civelek, Tuğçe Erkılıç; Uğur, DenizThe growing prevalence of visual disinformation has become an important problem to solve nowadays. Cheapfake is a new term used for the altered media generated by non-AI techniques. In their recent COSMOS work, the authors developed a self-supervised training strategy that detected whether different captions for a given image were out-of-context, meaning that even though pointing to the same object(s) in the image, the captions implied different meanings. In this paper, we propose four methods to improve the detection accuracy of COSMOS. These methods range from differential sensing and fake-or-fact checking that detect contradicting or fake captions to object-caption matching and threshold adjustment that modify the baseline algorithm for improved accuracy.ArticlePublication Metadata only Extending static code analysis with application-specific rules by analyzing runtime execution traces(Springer International Publishing, 2016) Ersoy, E.; Sözer, Hasan; Computer Science; SÖZER, HasanStatic analysis tools cannot detect violations of application-specific rules. They can be extended with specialized checkers that implement the verification of these rules. However, such rules are usually not documented explicitly. Moreover, the implementation of specialized checkers is a manual process that requires expertise. In this work, application-specific programming rules are automatically extracted from execution traces collected at runtime. These traces are analyzed offline to identify programming rules. Then, specialized checkers for these rules are introduced as extensions to a static analysis tool so that their violations can be checked throughout the source code. We implemented our approach for Java programs, considering 3 types of faults. We performed an evaluation with an industrial case study from the telecommunications domain. We were able to detect real faults with checkers that were generated based on the analysis of execution logs.ArticlePublication Metadata only A near Pareto optimal approach to student–supervisor allocation with two sided preferences and workload balance(Elsevier, 2019-03) Sanchez-Anguix, V.; Chalumuri, R.; Aydoğan, Reyhan; Julian, V.; Computer Science; AYDOĞAN, ReyhanThe problem of allocating students to supervisors for the development of a personal project or a dissertation is a crucial activity in the higher education environment, as it enables students to get feedback on their work from an expert and improve their personal, academic, and professional abilities. In this article, we propose a multi-objective and near Pareto optimal genetic algorithm for the allocation of students to supervisors. The allocation takes into consideration the students and supervisors' preferences on research/project topics, the lower and upper supervision quotas of supervisors, as well as the workload balance amongst supervisors. We introduce novel mutation and crossover operators for the studentsupervisor allocation problem. The experiments carried out show that the components of the genetic algorithm are more apt for the problem than classic components, and that the genetic algorithm is capable of producing allocations that are near Pareto optimal in a reasonable time.Conference paperPublication Metadata only FUSE-BEE: Fusion of subjective opinions through behavior estimation(IEEE, 2015) Şensoy, Murat; Kaplan, L.; Ayci, Gönül; de Mel, G.; Computer Science; ŞENSOY, Murat; Ayci, GönülInformation is critical in almost all decision making processes. Therefore, it is important to get the right information at the right time from the right sources. However, information sources may behave differently while providing information - i.e., they may provide unreliable, erroneous, noisy, or misleading information deliberately or unintentionally. Motivated by this observation, in this paper, we propose a statistical information fusion approach based on behavior estimation. Our approach transforms the conveyed information into more useful form by tempering them with the estimated behaviors of sources. Through extensive simulations, we have shown that our approach has a lower computational complexity, and achieves significantly low behavior estimation and fusion errors.ArticlePublication Metadata only Automatically learning usage behavior and generating event sequences for black-box testing of reactive systems(The ACM Digital Library, 2019-06) Kıraç, Mustafa Furkan; Aktemur, Tankut Barış; Sözer, Hasan; Gebizli, C. Ş.; Computer Science; KIRAÇ, Mustafa Furkan; AKTEMUR, Tankut Bariş; SÖZER, HasanWe propose a novel technique based on recurrent artificial neural networks to generate test cases for black-box testing of reactive systems. We combine functional testing inputs that are automatically generated from a model together with manually-applied test cases for robustness testing. We use this combination to train a long short-term memory (LSTM) network. As a result, the network learns an implicit representation of the usage behavior that is liable to failures. We use this network to generate new event sequences as test cases. We applied our approach in the context of an industrial case study for the black-box testing of a digital TV system. LSTM-generated test cases were able to reveal several faults, including critical ones, that were not detected with existing automated or manual testing activities. Our approach is complementary to model-based and exploratory testing, and the combined approach outperforms random testing in terms of both fault coverage and execution time.Conference paperPublication Metadata only Dağıtık bir openflow kontrol birimi mimarisi(IEEE, 2012) Yazıcı, Volkan; Sunay, Mehmet Oğuz; Ercan, Ali Özer; Electrical & Electronics Engineering; Computer Science; ERCAN, Ali Özer; SUNAY, Mehmet Oğuz; YAZICI, VolkanConsidering the modern internet traffic rates, the network architecture is of particular importance as the running services itself. On the other hand, due to the increasing complexity and black-box structure of the available networking hardware (switches, routers, etc.), the necessary network innovation imposed by the running services becomes infeasible in practice. The software-defined networking notion introduced to solve this problem and one of its emerging and powerful implementations, the OpenFlow protocol, advocate the idea of providing the control and data paths in separate planes. A network operating system running on this control plane, is anticipated to provide necessary measures for scalability and reliability in order to stand against the gigantic traffic pumped by the network. In this paper, we propose a distributed OpenFlow network operating system built with necessary scalability and reliability qualifications without requiring any changes to the existing OpenFlow protocol and networking equipment.ArticlePublication Metadata only A longitudinal case study on Nexus transformation: Impact on productivity, quality, and motivation(Wiley, 2023-09) Ersoy, E.; Çallı, E.; Erdoğan, B.; Bağrıyanık, S.; Sözer, Hasan; Computer Science; SÖZER, HasanThere have been success stories reported regarding the adoption of agile software development methods in the industry. There also exist observations on their limitations. One of these limitations is scalability since agile methods like Scrum were originally designed for small software teams. Scalable agile frameworks were introduced to address this limitation. We conducted an industrial case study on the adoption of such a framework, called Nexus. Our study involves quantitative and qualitative evaluation based on observations within a product development organization over a period of 12 months. Scrum is used for the development of a product during the first 6 months of this period. Nexus is used in the remaining 6 months. Data are collected throughout the whole period for measuring productivity, quality, and team member motivation. Results suggest a significant increase in productivity and product quality after switching to Nexus. Team motivation was slightly improved as well.