139585 research outputs found
Sort by
Scoping review on the economic aspects of machine learning applications in healthcare
No full text| openaire: EC/H2020/101016775/EU//INTERVENEBackground: The development and use of artificial intelligence and machine learning technologies in healthcare have increased, prompting a need for evidence on their safety and value. Economic evaluations support healthcare decision-making and resource allocation. This scoping review aimed to map and synthesize current approaches to evaluating the economic aspects of machine learning based technologies implemented in healthcare. Methods: Following the updated JBI guidance for scoping reviews, six databases (PubMed, CINAHL, Cochrane Library, Embase, Scopus, and IEEE Xplore) were searched for studies evaluating the economic aspects of machine learning-based technologies within healthcare. No exclusions were applied to healthcare settings, healthcare professionals or used economic evaluation methods. The results of data extraction were analyzed using descriptive statistics and inductive coding. The reporting of the studies was compared against the CHEERS-AI statement. Results: A total of 6332 references were retrieved, with 18 studies included in the review. The studies comprised economic evaluations (n = 9), impact evaluations (n = 5), and performance evaluations (n = 4), with cost-effectiveness analysis being the most frequently used economic evaluation method (n = 8). The comparison of the studies to the reporting guidelines revealed gaps in the reporting of details from economic evaluations and the artificial intelligence nature of the technologies. Overall, the study alignment with the CHEERS-AI items on average was 39.6 %, with 64.1 % alignment with economic evaluation details, and 21.3 % alignment with key details related to the artificial intelligence nature of the evaluated technologies. Conclusions: The current literature evaluating the economic aspects of machine learning-based technologies implemented in healthcare reveals gaps in coherence and coverage. Frameworks guiding artificial intelligence development should be refined to incorporate components related to system evaluation and post-implementation considerations. Further, multidisciplinary collaboration should be enhanced and promoted.Peer reviewe
Advances in robust signal processing and applications
No full textRobust signal processing and machine learning methodologies are critical for the reliable operation of modern technological systems, particularly in dynamic and uncertain environments such as the Internet of Things (IoT).
However, system performance is often compromised by pervasive challenges, including structural perturbations in graph-based models, complex non-Gaussian noise in communication systems, and the structural heterogeneity of high-dimensional tensor data. This thesis addresses these critical challenges by developing a suite of robust methodologies grounded in distinct yet complementary perspectives on robustness.
First, this research establishes a comprehensive analytical framework to quantify the sensitivity of Graph Convolutional Neural Networks (GCNNs) to probabilistic graph perturbations. Tight, expected bounds for Graph Shift Operator (GSO) errors are derived without requiring eigendecomposition, and a linear relationship between GSO perturbations and GCNN output differences is revealed, providing theoretical stability guarantees for multilayer architectures.
Second, novel robust device activity detection (AD) algorithms are developed for massive random access systems operating under challenging non-Gaussian noise. By formulating AD objectives based on robust loss functions (e.g., Huber's loss) and proving the geodesic convexity of the conditional objective, efficient fixed-point, coordinate-wise, and matching pursuit algorithms with proven convergence are proposed. These methods significantly outperform traditional Gaussianbased approaches in heavy-tailed noise environments.
Third, a generalized Nonnegative Structured Kruskal Tensor Regression (NS-KTR) framework is introduced for the effective and interpretable modeling of high-dimensional tensor data. This framework integrates non-negativity constraints with mode-specific hybrid regularizations (e.g., LASSO, total variation, ridge), accommodates both linear and logistic regression, and is solved via an efficient ADMM-based algorithm.
Collectively, this thesis advances the theory and practice of robust signal processing by providing novel tools for ensuring stability, resilience to distributional deviations, and robust modeling through structural priors. The developed frameworks and algorithms contribute to the design of more reliable and efficient signal processing systems for real-world applications.navigointi mahdollistakuvilla vaihtoehtoiset kuvauksettaulukot saavutettaviastrukturell navigationalternativa textuella beskrivningar för bildertabeller tillgängligastructural navigationalternative textual descriptions for imagestables accessibl
Building a global business school in the local language? : A knotted and nested paradox system in shaping international student inclusion
No full textPublisher Copyright: © Academy of Management Learning & Education.In this paper, we offer a study of a global business school in a university located in a non-English-speaking country. We focus on paradoxical tensions related to the use of English as the global language of higher education, and the use of the local language as a mirror of national interests. Our study shows how underlying global–local tensions manifest as a nested linguistic paradox of universalism–particularism across international and national, organizational, and group levels. This paradox is knotted to other tensions on each level, which are made salient by either–or responses to the linguistic paradox. We contribute to paradox theory by theorizing a nested and knotted paradox system, by explaining how either–or responses cause pole suppression and disempowerment that cascade across levels, and by elaborating on how paradox and power operate in language use. We also advance research on inclusion in business schools by theorizing international students’ experiences of inclusion from a paradox and language perspective. Our study indicates that in the era of rising nationalism around the world, the resurgence of local languages in business schools and universities and its consequences for student inclusion is a timely subject of inquiry.Peer reviewe
A probabilistic-driven approach for early design quality risk and crux identification using non-Markovian stochastic Petri nets
No full textPublisher Copyright: © The Author(s) 2025.Quality risk analysis of high-process-oriented systems, which refers to their ability to achieve required tasks on time, receives little attention during the early conceptual design stage, primarily due to the high level of abstraction when the system form is not yet fully defined. Although several mathematical methods exist to address this issue, they are fragmented across domains and lack a unified integration into early design practice. To address this problem, this paper introduces a novel approach that models design problems as discrete events with output conflict representation, using the non-Markovian stochastic Petri net. The framework is further integrated with mathematical techniques, including semi-Markov performance evaluation, sensitivity analysis, and uncertainty analysis, to quantify quality risks and identify the design crux (the most critical design parameters). By incorporating Monte Carlo simulations, it facilitates designers and engineers with early insights and allows them to compare alternative design specifications. Its applicability is demonstrated through a case study on the conceptual development of a remote maintenance system for the In-Bioshield area of the EU-DEMO fusion power plant. Initial results showed potential in identifying quality risks, addressing key factors contributing to the design problem, and finding optimal design specifications in the early stages.Peer reviewe
Carrier mobility in crystalline germanium at high injection: experimental characterization of carrier-carrier scattering
No full textThe decay of the sum of electron and hole mobilities, μs = μn+μp, due to carrier-carrier scattering was experimentally investigated in crystalline germanium (Ge) at high-injection conditions. Contactless measurements of the mobility sum as a function of the excess carrier density (Δn) in Ge were obtained using photoconductance decay methods. First, the measurement method was revised and improvements were introduced to ensure that μs(Δn) could be obtained for independent samples with improved accuracy. This method is successfully validated with crystalline silicon and, then, applied to Ge samples of different doping types and resistivity. The analysis of the data suggests that the mobility decay at high injection levels cannot be properly explained with the usual assumption of equal cross section for carrier-carrier and carrier-ion scattering events. Instead, we find the mobility sum due to carrier-carrier scattering to be inversely proportional to Δn according to the expression 8 × 1020·Δn−1 cm2V−1s−1. The limitations and potential error sources of the measurement method are discussed and, finally, the mobility model is used to improve lifetime analysis at high injection, allowing to estimate the ambipolar Auger recombination coefficient at Camb = 7 × 10−31 cm6s−1.Peer reviewe
Generalized nonnegative structured Kruskal tensor regression
No full textPublisher Copyright: © 2025 The AuthorsThis paper introduces Generalized Nonnegative Structured Kruskal Tensor Regression (NS-KTR), a novel tensor regression framework that enhances interpretability and performance through mode-specific hybrid regularization and nonnegativity constraints. Our approach accommodates both linear and logistic regression formulations for diverse response variables while addressing the structural heterogeneity inherent in multidimensional tensor data. We integrate fused LASSO, total variation, and ridge regularizers — each tailored to specific tensor modes — and develop an efficient alternating direction method of multipliers (ADMM)-based algorithm for parameter estimation. Comprehensive experiments on synthetic signals and real hyperspectral datasets demonstrate that NS-KTR consistently outperforms conventional tensor regression methods. The framework's ability to preserve distinct structural characteristics across tensor dimensions while ensuring physical interpretability makes it especially suitable for applications in signal processing and hyperspectral image analysis.Peer reviewe
A multicriteria modelling framework for evaluating clean energy transitions : the case of Greece as electricity exporter
No full textPublisher Copyright: © 2025 The AuthorsThe development of clean power generation interacts with various policy objectives pertaining to the energy system, the general economy, people's health, critical resources, and biodiversity, which can be synergistic or conflicting. In this respect, assessing these synergies and trade-offs together within a multicriteria framework has become critical, yet remains inadequately addressed in the literature. This paper contributes to the literature by examining a broad spectrum of trade-offs associated with clean energy development in a Mediterranean country, namely Greece, along with the associated economic impacts of it becoming a clean electricity supplier to its neighbouring countries. In this context, the study further contributes to the literature by developing and linking a technology-rich optimization model representing Greece's energy and resource systems with a Recursive-Dynamic Computable General Equilibrium model for the Greek economy. These models are coupled with a multicriteria framework, employing the hybrid AHP-TOPSIS method along with proxy indicators reflecting impacts on costs, critical resources, energy security, people's health, biodiversity, and the economy. The results revealed that policies involving electricity exports stand out as “no regret” options for most decision-maker types, potentially expanding the Greek economy by an average of 0.12 %–0.26 % annually over the 2025–2050 period, depending on electricity price evolution, compared to a baseline scenario in which the country remains a net electricity importer throughout the entire period.Peer reviewe
Queueing Analysis of an Ensemble Machine Learning System
No full textPublisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.Recent advances in AI/ML technologies have accelerated the development of various ML applications. One of the major trends in AI/ML application development is the increasing use of multiple ML models to support high-accuracy inference in a complex end-to-end ML serving. However, testing the right configuration of multiple ML models is expensive, and the application requirements for ML inferences are highly dependent on various factors like the quality of ML models, computing resource performance, and data quality. In this context, techniques and methods that help to emulate and analyze ML inference characteristics using queueing theory can reduce the development effort and cost for ML services encapsulating ML models but also the entire ML system. In this paper, we modeled and analyzed a queueing model for an ML system that uses ensemble learning as an inference method with a new rule and clarified the impacts of model design in ensemble learning on the system’s performance. As a result, we demonstrate the usefulness of the analysis for understanding possible configurations and their efficiency in the ML system through queueing analysis and simulation.Peer reviewe
Machine learning-assisted development of polypyrrole-grafted yarns for e-textiles
No full textPublisher Copyright: © 2024 The AuthorsThe development of digitally enhanced fabrics is growing, but progress is currently being hampered by a lack of sustainable alternatives to metallic conductors. In particular, the process of testing and optimizing new candidate materials is both time-consuming and resource intensive. To address these challenges, we present a machine learning-assisted approach to the design of fully-textile based conductive e-textile prototypes. Based on commercially available Tencel yarn coated with polypyrrole, with 11 experiments we were able to establish the global optimum of the reaction and estimate the noise, crucial for the understanding of the electrical resistance's behavior. The reaction conditions are optimized for conductivity and cost-effectiveness by means of Bayesian optimization and Pareto front analysis. Notably, we find that the addition of p-toluenesulfonic acid as a dopant does not significantly influence the conductivity of the yarn and provide a possible rationale based on the surface morphology of the yarn. The optimized yarns are woven into prototype fabrics with different patterns, and we demonstrate their applicability as flexible conductive wearable and heaters.Peer reviewe
Viheliäiset ongelmat hyvinvointivaltiossa: Segregaatio kaupunkipoltiikassa
No full textSegregation is a relatively new research topic in Finland. Levels of residential segregation are growing, presenting a problem for the welfare state ethos. This thesis investigates how urbanplanning policy deals with the 'problem' of segregation in Finland. The four articles study the interrelationship between the social, the physical, and the perceived city and each dimension's role in the segregation cycle. The first three articles concentrate on the Helsinki metropolitan area, analyzing the social dimension of institutionalized urban policies such as transit-oriented development and social mixing. The last article examines how segregation is recognized in the twenty largest Finnish cities.
Perceptions are of interest to urban planning, as they are linked to selective moving patterns, which are one driver of the segregation process. The first article finds interlinkages between neighborhood satisfaction, socioeconomic status, and the share of social housing in neighborhoods. The second article finds differences in neighborhood satisfaction by tenure status, with municipal tenants reporting lower neighborhood satisfaction, quality of life, and perceived safety than homeowners. The second article concludes that while social mixing seems to have bridged the gap in spatial justice among different tenure groups, it has not managed to equalize neighborhood perceptions in the Helsinki metropolitan area.
The third article concludes that while the segregation trajectory in the Helsinki metropolitan area is perceived as alarming and needing intervention, governance capacity is lacking: segregation is poorly articulated and yet to be institutionalized. The fourth article concludes that acknowledging segregation depends on city size and urban policy framework. Where segregation is named as a goal, it is often not translated into explicit actions in local policies. Segregation is mostly targeted with housing and land use policies, cornerstones of the local autonomy. This model works poorly in a situation where segregation is a regional issue. Governance capacity is also lacking on the state level, where housing policies affecting segregation are volatile. Insufficient governance capacity carries a risk: spatial inequalities may eventually become structural and cemented.Segregaatio on suhteellinen uusi tutkimusaihe Suomessa. Segregaatio on kuitenkin voimistunut 2000-luvulla kaikissa Suomen suurimmissa kaupungeissa haastaen hyvinvointivaltion eetoksen. Väitöskirjan neljä artikkelia tutkivat sosiaalisen, fyysisen ja koetun kaupungin välistä vuorovaikutusta sekä kunkin ulottuvuuden roolia segregaatioprosessissa. Kolme ensimmäistä artikkelia keskittyvät pääkaupunkiseutuun ja analysoivat segregaation ja saavutettavuuden sekä sosiaalisen sekoittamisen yhteyttä. Viimeinen artikkeli kysyy, miten segregaatio tunnistetaan Suomen suurimmissa kaupungeissa.
Asukaskokemukset ovat yhteydessä valikoivaan muuttoliikkeeseen, joka puolestaan on yksi segregaatioprosessin ajureista. Pääkaupunkiseudulla on yhteys asuinaluetyytyväisyyden, asuinalueen sosioekonomisen aseman sekä ARA-vuokra-asuntojen osuuden välillä. Asuinaluetyytyväisyys vaihtelee hallintamuodon mukaan: kaupunkien ARA-vuokra-asukkaat kokevat asuinaluetyytyväisyytensä, elämänlaatunsa ja alueensa turvallisuuden huonompina kuin omistusasunnoissa asuvat, vaikka objektiivisesti mitattuna eroja esim. palveluiden sijainnissa ei ole hallintamuotojen välillä. Sosiaalinen sekoittaminen näyttää siis onnistuneen paremmin tasaarvoisen palveluverkon luomisessa kuin subjektiivisten kokemusten tasaamisessa hallintamuotojen välillä.
Segregaation kehityssuunta koetaan huolestuttavana ja toimenpiteitä vaativana. Segregaatio on kuitenkin huonosti artikuloitu kaupunkipolitiikassa, vaikkakin esiintyy pääkaupunkiseudun kaupunkien, seudun ja valtion ylätavoitteena. Viimeinen artikkeli toteaa, että segregaation tunnistaminen riippuu kaupungin koosta ja kaupunkipolitiikan ’kivijalasta’. Segregaation maininta ylätavoitteena ei takaa ilmiön kanavoitumista kaupunkipoliittisiin ohjelmiin eksplisiittisinä toimenpiteinä. Lisäksi segregaatiotoimenpiteet rajoittuvat yksittäisiin kaupunkeihin, vaikka ilmiö on seudullinen. Valtion ailahteleva asuntopolitiikka on lisäksi vaikeuttanut segregaation torjuntaa. Kun mikään hallinon taso ei saa otetta kehitykseen, segregaatio uhkaa muuntua sosio-spatiaalisesta ongelmasta vaikeammin ratkaistavissa olevaksi, rakenteelliseksi ongelmaksi