5916 research outputs found
Sort by
Affective polarization and democratic backsliding
No full textEvidence is strong that affective polarization and democratic backsliding are interlinked phenomena. But research still examines the causal directions between the two, mediating causal mechanisms, and the relation between individual-level polarization, norms, and attitudes and group-level political behavior. Highlighting potential ecological inference issues underlying individual-level inquiries, we focus on the systemic-level links between affective polarization and backsliding. While mass affective polarization may or may not contribute to other modes of autocratization, it is especially likely to facilitate democratic backsliding because it occurs under popularly elected governments and gradually. Investigating the polity-level and qualitative dynamics of polarization as a process is necessary to understand many causal links between affective polarization and backsliding. Crucial are, for example, political elite interests, framing and agency, interactive group dynamics, and discursive trends and institutional incentives. The impact of group-level affective polarization on democratic outcomes may evolve through discontinuous leaps, which quantitative operationalizations may only partially be able to capture. The impact can be disproportionately negative once polarization settles on a pernicious state. Different types and levels of polarization coexist, reinforce, and follow each other in chain reactions, leading to a multilayered process.Publisher versio
Analytical assessment of optimal number and placement of sensors for modal response estimation of frame buildings
No full textContinuous monitoring of the structural health of buildings can be achieved by placing acceleration sensors at various locations on the structure. Considerable variations in the dynamic response captured through these measurements can be used to identify damage due to aging or extreme load effects. Accurate estimation of modal dynamic behavior using acceleration measurements is critical for efficient structural health monitoring (SHM) of buildings under limited economic resources. In this study, an analytical approach is presented to determine theoretically the optimal number and placement of sensors in a typical five-story reinforced concrete frame building, ensuring that accuracy in modal response is not compromised. In this regard, acceleration responses at each floor level are recorded from time-history analyses of the building’s finite element model under several earthquake ground motions, simulating a complete set of SHM system measurements during seismic events. Then, considering exhaustive search optimization for sensor placement, the operational modal analysis method of Enhanced Frequency Domain Decomposition (EFDD) is applied to extract modal parameters such as the first-mode natural frequency, damping ratio, and mode shape from the simulated monitoring data. The accuracy of the estimated modal parameters from different sensor design alternatives is compared with those from the benchmark sensor configuration to determine the optimal number and locations of sensors. The findings of this research also highlight the importance of using different modal identification methods in SHM. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025
Financial asset price prediction with graph neural network-based temporal deep learning models
No full textChanges in the prices of multiple financial assets over time can be characterized by their complex nature and interdependence. More traditional forecasting approaches may overlook the interdependencies among these assets, since they may not fully consider the spatial-temporal dependencies between them. Graph neural networks (GNNs) have emerged as powerful tools for modeling complex relational dependencies in areas such as social network analysis and traffic forecasting. However, their application in asset price prediction remains relatively unexplored. Here, we investigate GNNs’ effectiveness in forecasting multiple financial asset prices jointly, specifically in the foreign exchange (Forex) and cryptocurrency markets. We employ three spatio-temporal GNN frameworks-MTGNN, StemGNN, and FourierGNN-which are all recognized for their state-of-the-art performance in forecasting multivariate time series. These models transform time-series data into graphs and capture both spatial and temporal dependencies. They significantly outperform the baseline methods, including LSTM, ARIMA, and VAR, in predicting financial asset prices in the highly volatile cryptocurrency market. While the performance gap is less obvious in the relatively stable Forex market, GNN-based models still demonstrate a general advantage over LSTM, although they are outperformed by ARIMA. Through a series of experiments and backtesting strategies, we assess the predictive power and profitability of these models in portfolio construction. Our code and datasets are publicly available at https://github.com/seferlab/temporal gnn
Feline events beyond pet ownership paradigm: An ethnographic case report on end-of-life decision-making for NONAME 6796
No full textBackground: The human-animal bond is most often understood within the framework of the "pet ownership paradigm," which defines a permanent relationship between an animal and their legal guardian. However, this framework is insufficient to explain the profound, yet temporary bonds formed with unowned urban animals, particularly during medical crises. This case report explores a critical incident that necessitated human intervention and challenged existing ethical frameworks in veterinary palliative care, along with the social conditions surrounding it. Methods: This study adopts a Constructivist Grounded Theory approach and collects ethnographic data through participant observation. Following the discovery of a kitten with spinal cord injury in Istanbul, the researcher observed the ensuing events in the field and documented them through photographs and field notes. The data were analyzed using ATLAS.ti qualitative data analysis software, with a focus on the functioning of veterinary systems, the dynamics of temporary care networks, and the experiences of human proxies in the field. Results: This encounter, conceptualized as a "feline event," transformed ordinary bystanders into "human proxies" without any legal ownership status. Interactions with multiple veterinary clinics revealed a system structured around ownership, from administrative registration to treatment protocols. The kitten's registration under the code "NONAME 6796" disclosed both the individuality of the feline patient and the institutional anonymity faced by animals outside the "pet ownership paradigm" when no legal guardian is present. Human proxies within the temporary care network were forced to choose between an expensive and uncertain surgery that required permanent guardianship, and euthanasia, leading to a profound ethical dilemma. The decision to proceed with euthanasia, based on a poor prognosis and limited resources, highlighted that end-of-life processes for stray animals are shaped not only by medical but also by social factors. Conclusion: The case of NONAME 6796 demonstrates that powerful human-animal bonds and significant care responsibilities can exist beyond the boundaries of legal ownership. Veterinary ethics and urban care systems need to be expanded to support spontaneous care networks that emerge to help animals in need. This requires designing "more-than-human infrastructures of care" that can accommodate these temporary yet meaningful relationships, facilitate urban animals' access to veterinary care, and alleviate the ethical dilemmas faced by temporary caregivers.Publisher versio
Seismic behaviour of multistory prefabricated modular steel buildings in high seismic regions
No full textModular steel building systems provide effective solutions for temporary and permanent accommodation, particularly in critical situations such as post-earthquake scenarios. These systems address urgent needs for rapid construction while maintaining high-quality standards, making them essential in modern construction. By staggering multiple master modular units both horizontally and vertically, diverse architectural layouts are created to meet specific requirements. This study evaluates the seismic performance of a prototype five story modular building using 3D finite element analysis combined with nonlinear time-history analysis. The finite element model, which accounts for material and geometric nonlinearities, is used to monitor building drift, material yielding, plasticity, and anchorage reactions. Typical interconnection details are provided, and the seismic demands for these connections are assessed
Hypergraph neural networks to predict stock movements by exploring higher-order relationships
No full textPredicting stock price movements can be framed as a classification task, where the goal is to anticipate whether a stock will increase, decrease, or remain stable. Most existing approaches rely solely on the movement patterns of individual stocks or stock pairs, overlooking the more complex, higher-order connections that exist among groups of stocks. In practice, stocks are often interrelated in higher orders, for example, by belonging to the same industry sector or being jointly held within the same investment fund. To address this, we compare 4 hypergraph neural network-based approaches to make spatio-temporal predictions for stock movement prediction, which explicitly leverages these higher-order dependencies. We use two heterogeneous hypergraphs, where one hypergraph represents sector-based associations and the other one represents fund-holding relationships among stocks. In general, we found the hierarchical hypergraph attention mechanism and temporal attention to be effective in achieving better performance. A hierarchical hypergraph attention mechanism models these relationships by weighting the contributions of stock nodes, hyperedges, and even the hypergraphs themselves. Temporal attention captures time-dependent dynamics of both stock and sector sequences, effectively accounting for the influence of past states. Experiments on real-world datasets demonstrate that the methods specializing in hypergraph integration achieve superior performance compared to existing methods, both in terms of predictive accuracy and profitability.Publisher versio
Providing edge to cloud continuum with adaptive model selection and operational score
No full textAdvancements in Deep Neural Network (DNN) models and hardware accelerators have made edge intelligence a practical alternative to cloud-based intelligence. However, applicationspecific requirements, such as accuracy, latency, security, and privacy, as well as workload fluctuations, necessitate dynamic allocation of edge and cloud resources. To facilitate such dynamic allocation, we propose an adaptive model selection and switching framework that leverages operational performance scores. We evaluate the approach using various object classification models, demonstrating its ability to balance accuracy and inference time while ensuring scalability and efficient resource utilization.Norges Forskningsråd ; European Commission ; European Lighthouse To Manifest Trustworthy and Green AI ; EU Horizon Research and Innovation Program ; SF
Ultrafast laser synthesis of zeolites
No full textResearch demonstrates that zeolite nucleation and growth can be controlled by fine-tuning chemical composition, temperature, and pressure, resulting in structures with diverse porosities and functionalities. Nevertheless, current energy delivery methods lack the finesse required to operate on the femto- and picosecond timescales of silica polymerization and depolymerization, limiting their ability to direct synthesis with high precision. To overcome this limitation, an ultrafast laser synthesis technique is introduced, capable of delivering energy at these timescales with unprecedented spatiotemporal precision. Unlike conventional or emerging approaches, this method bypasses the need for specific temperature and pressure settings, as nucleation and growth are governed by dynamic phenomena arising from nonlinear light-matter interactions, such as convective flows, cavitation bubbles, plasma formation, and shock waves. These processes can be initiated, paused, and resumed within fractions of a second, effectively "freezing" structures at any stage of self-assembly. Using this approach, the entire nucleation and growth pathway of laser-synthesized TPA-silicate-1 zeolites is traced, from early oligomer formation to fully developed crystals. The unprecedented spatiotemporal control of this technique unlocks new avenues for manipulating reaction pathways and exploring the vast configurational space of zeolites.European Research Council (ERC) ; TÜBİTAKPublisher versio