256869 research outputs found
Sort by
Infants relax in response to unfamiliar foreign lullabies
No full textMusic is characterized by acoustical forms that are predictive of its behavioral functions. For example, adult listeners accurately identify unfamiliar lullabies as infant-directed on the basis of their musical features alone. This property could reflect a function of listeners’ experiences, the basic design of the human mind, or both. Here, we show that American infants (N = 144) relax in response to 8 unfamiliar foreign lullabies, relative to matched non-lullaby songs from other foreign societies, as indexed by heart rate, pupillometry, and electrodermal activity. They do so consistently throughout the first year of life, suggesting the response is not a function of their musical experiences, which are limited relative to those of adults. The infants’ parents overwhelmingly chose lullabies as the songs that they themselves would use to calm their fussy infant, despite their unfamiliarity. Together, these findings suggest that infants are predisposed to respond to universal features of lullabies
Architect, Activist, Immigrant, Mother: Edith Schreiber-Aujame and the Potential of Female Architectural Agency
No full textThe work of Edith Schreiber-Aujame has been framed as derivative of or inspired by a host of masculine influences; however, Schreiber-Aujame forged a strong identity and successful career apart from male partners, employers, and colleagues, while her experiences as a woman, immigrant, and mother shaped her professional ambitions and personal politics of care. Following introduction to Schreiber-Aujame, the woman, this chapter addresses her creation of the planning and pedagogy bureau, l’Association pour l’Environnement Pédagogique (AEP). Led by female architects and educators, AEP’s output has been overlooked as the domain of women’s design, while its collaborative methods challenge dominant notions of genius and authorship. Revisiting the life and career of Schreiber-Aujame illustrates the promises and pitfalls of a biographical approach in attempts to historicise, reclaim, and solidify female architectural agency
Effect of Fe addition on the ORR electrocatalytic activity of precursor‐derived high‐entropy carbide nanocomposites
No full textHigh-entropy carbides (HECs) exhibit enhanced mechanical properties and high temperature oxidation resistance as compared to their binary counterparts. However, their application as catalytic materials was restricted due to a lack of high specific surface area with conventional synthesis methods such as solid-state sintering. In the present work, porous high-entropy carbide nanocomposites with varying elemental composition were synthesized through the precursor-derived ceramic route, and their electrocatalytic activity for the oxygen reduction reaction (ORR) was assessed. Thus, the compositionally complex nanocomposite was prepared from a preceramic precursor, which was, upon modification of a polysiloxane with Ta, V, W, Mo, and Nb-based organometallic precursors via pyrolysis in an inert atmosphere. The pyrolyzed ceramic was subsequently modified with Fe-based precursor, with varying quantities and heat-treated at 1300°C to obtain Fe-containing high-entropy carbide nanocomposites. X-ray diffraction and ADF-STEM imaging coupled with EDX confirmed the incorporation of Fe into the HEC lattice. The onset potential for the electrocatalytic ORR increased from 0.81 to 0.85 V upon incorporation of Fe into the high-entropy carbide nanocomposite. Furthermore, with the addition of electron-rich Fe in HEC nanocomposite, the reaction mechanism moved toward a four-electron path. This suggests the positive effects of incorporating Fe into HEC nanocomposites
High Voltage Robustness of Single Phase Immersion Cooled Power Modules Considering Electric Aviation Requirements
No full textEin systematischer Vergleich von Named Entity Recognition Ansätzen für Cyber Threat Intelligence
No full textCyber Threat Intelligence (CTI) ist ein essenzieller Teil der Cybersecurity. Sie umfasst Wissen, das Technologien und Individuen zur Abwehr von Cyberangriffen benötigen. Angesichts der großen Menge an täglich generierten CTI-Informationen, benötigen Cybersicherheits-Analysten automatisierte Werkzeuge, um diese Daten effizient zu verarbeiten, um aussagekräftige Erkenntnisse zu erlangen und davon Handlungsweisen abzuleiten. Eines solcher Werkzeuge ist Named Entity Recognition (NER). NER identifiziert relevante Entitäten in unstrukturiertem Text, um Folgeaufgaben, wie Suche, Filterung und Analyse zu erleichtern. In dieser Arbeit werden Ansätze für NER auf CTI-Daten im Rahmen einer Systematic Literature Review (SLR) untersucht, die mithilfe eines in dieser Studie vorgeschlagenen LLM-basierten Extraktors durchgeführt wird, und im Hinblick auf ihre Architekturen analysiert. Darüber hinaus werden sechs Publikationen, die ihre Ansätze öffentlich bereitstellen, hinsichtlich ihrer Tagging-Performance sowie ihrer Trainings- und Betriebskosten evaluiert und miteinander verglichen. Um aussagekräftige Einblicke in die Tagging-Performance zu ermöglichen, wird diese aus vier Perspektiven analysiert: der Gesamtperformance, der Performance auf Entitätstypen, der Performance in Bezug auf Entitätsattribute (z.B. Entitätshäufigkeit) sowie der Fähigkeit, vollständige Dokumente, anstatt einzelne Sätze, zu taggen. Dies ermöglicht eine fundierte Entscheidungsfindung bei der Auswahl des am besten geeigneten NER-Ansatzes für das Tagging von CTI-Nachrichten. Die Evaluation zeigt, dass einfachere Architekturen über alle analysierten Perspektiven hinweg konsistent bessere Ergebnisse erzielen als komplexere Alternativen. Encoder-basierte Language Models (LMs) wie BERT, kombiniert mit einer einfachen Decoder-Ebene (CRF oder Softmax), erreichen die höchste Performance. Entgegen den Erwartungen führt die Einbeziehung von Kontext auf Dokumentenebene nicht zu einer Leistungsverbesserung, sondern verschlechtert die Ergebnisse. Die Analyse zeigt ferner, dass in den meisten betrachteten Veröffentlichungen NER-Performancewerte künstlich erhöht sind, verursacht durch Entitätsüberlappungen zwischen Trainings- und Testdaten. Prompt-basierte Extraktoren, die auf dem allgemeinen Wissen großer Large Language Models (LLMs) beruhen, schneiden nicht nur schlechter ab als domänenspezifische Ansätze, sondern verursachen auch deutlich höhere Kosten, mit bis zu 50-fach langsamerer Inferenz und über 100-fach höherem Energieverbrauch, was sie für CTI-Anwendungen mit hohem Durchsatz ungeeignet macht.Cyber Threat Intelligence (CTI) is an essential element of the cybersecurity domain. It encompasses knowledge that supports technologies and individuals in mitigating cyber attacks. Given the large volume of cybersecurity information generated daily, cybersecurity analysts require automated tools to efficiently process this data and extract meaningful insights and actionable outcomes. One such concept is Named Entity Recognition (NER), which identifies and labels relevant entities in text to improve searching, filtering and analysis. In this work, approaches for NER on CTI data are examined through a Systematic Literature Review (SLR) enabled by an LLM-based extractor proposed in this study and analyzed with respect to their architectures. Furthermore, six publications that publicly provide their approaches are evaluated and compared based on tagging performance as well as training and operational costs. To provide meaningful insights into tagging performance, we analyze it from four perspectives: overall performance, performance on specific entity types, performance on specific entity attributes (e.g. entity frequency) and the capability to tag entire documents. This enables informed decision-making in selecting the most suitable NER approach for tagging CTI news. The evaluation shows that simpler architectures consistently outperform more complex alternatives across all analyzed perspectives. Encoder-based language models (LMs) such as BERT, combined with non-complex decoding layers (CRF or Softmax), achieve the highest tagging performance. Contrary to expectations, the incorporation of document-level context does not improve performance, but degrade results. The analysis further indicates that NER evaluation performance scores are inflated due to data leakage, caused by entity overlap between training and test data. Prompt-based extractors that rely on the general knowledge of large language models (LLMs) not only underperform compared to domain-specific approaches but also incur substantially higher computational costs, with inference up to 50 times slower and over 100 times greater energy consumption, which renders them unsuitable for high-throughput CTI applications
A distributed-parameter observer for estimating the distribution of concentrations in a tubular protein refolding reactor
No full textInclusion bodies formed during recombinant protein expression in Escherichia coli serve as a high-yield source of target proteins but require refolding to regain their native structure. Continuous refolding processes, such as tubular refolding, provide advantages over batch and fed-batch methods by enabling precise control of key process variables like concentration, temperature and refolding time. In this contribution, a real-time monitoring framework based on an extended Kalman filter configuration is presented. This filter configuration enables accurate, dynamic concentration distribution estimation by fusing continuous inline measurements of refolded protein with periodic at-line measurements of aggregated species. The combination of both measurements enhances observer precision and ensures consistent product quality. To further improve the estimation, the observer scheme is augmented to simultaneously estimate reaction rates during operation. For determining the optimal sensor locations, the impact of the measurement positions along the tube on the accuracy of the state estimates is studied. The model-based extended Kalman filter utilizes a partial differential equation model that captures the protein refolding kinetics within a tubular reactor. Coupled with a reduced-order model, real-time capable and feasible state correction is possible. This hybrid monitoring strategy improves refolding efficiency, yield, and scalability for large-scale protein production applications. Using a simulated reality, the observer methodology is validated and compared to an open-loop simulation of the refolding process
Woven Rigidly Foldable T-Hedral Tubes Along Translational Surfaces
No full textRigidly foldable tubes, initially derived from mirroring and interconnecting corresponding edges of Miura-ori cells, possess bi-directional flat-foldability and exhibit out-of-plane stiffness due to their one-degree-of-freedom kinematic movement in their regular configuration. Various methods have been established for enhancing their structural complexity and stiffness, including edge and face connections, as well as zipping and interleaving, leading to the creation of advanced structures like foldable sandwiches and metamaterials. In this study, we introduce a novel family of cellular structures where orthogonally aligned rigid-foldable tubes are in a woven configuration. We show a computational process to achieve the tubular designs following a translational surface for any given weaving pattern (including existing interleaved tubes) by introducing the idea of flip-flop joints that can switch the up-and-down relation between contact tubes by changing their orientation. This research not only expands the repertoire of origami-based engineering techniques but also offers a practical toolset for designers and engineers seeking innovative solutions in structural optimization
Three-phase particle reinforced composites: Effective fields and the Mori–Tanaka method
No full textThe Mori–Tanaka method is a micromechanical effective-field model that approximates the local fields acting on any inhomogeneity phase by the corresponding matrix field. This assumption is tested in a case study using finite-element-based periodic homogenization for numerically evaluating the effective-field inhomogeneity concentration tensors of the reinforcement phases for a set of simple three-phase particle reinforced composites. The numerical predictions show a clear dependence of the mechanical and thermal effective fields on the material properties of the particulate phases. Even though this behavior deviates from the assumption underlying the Mori–Tanaka method, the latter provides useful approximations for the macroscopic stiffnesses and conductivities for the set of composites covered by the study
Low-Frequency Polarization of Blank Ice Features in Solid Rocks
No full textFor hydrogeological management of seasonally frozen soils, or permafrost, the quantification of ice and water content is key. Changes in electrical conductivity are commonly used to monitor ice-to-water ratios but hard rocks, ice and air are all highly resistive materials. The high-frequency polarization of ice (kHz range) offers a suitable alternative to the conductivity, but field investigations at such frequencies are still challenging. Here, we demonstrate the polarization of blank ice features in solid rocks at low frequencies (1–100 Hz). We identify two polarization effects: (a) an increase in the response due to the polarization of charges in the electrical double layer formed at the ice-water-rock interface; (b) an even larger response due to the non-equilibrium freezing potential taking place during ice formation. We demonstrate that these effects can be used to image ice features, for instance, during ground-ice formation and degradation
Enhancing the Sustainability of Steelmaking through the Optimization of Ladle Operations
Get PDFStahl ist ein Grundpfeiler der modernen Gesellschaft, doch seine Herstellung gehört nach wie vor zu den energie- und emissionsintensivsten industriellen Aktivitäten weltweit. Um die bis 2050 geforderten tiefgreifenden Dekarbonisierungsziele zu erreichen, sind innovative Optimierungsmethoden zur Verbesserung der Effizienz in der Stahlherstellung erforderlich. Diese Arbeit untersucht, wie das thermische Verhalten und die Abnutzung der feuerfesten Auskleidung von Stahlpfannen, einem wichtigen Behälter in der Sekundärmetallurgie, in Entscheidungshilfesysteme integriert werden können, um die Nachhaltigkeit des Pfannenbetriebs zu verbessern. Zu diesem Zweck wurden gemischt-ganzzahlige lineare Programmierung und datengesteuerte Vorhersagemodelle kombiniert, um die CO2-Emissionen im Zusammenhang mit der Einsatzplanung von Pfannen und der Wartungsplanung der Flotte zu minimieren und so optimale Betriebsabläufe zu ermitteln. Schließlich wurde ein Prototyp eines Entscheidungshilfesystems entwickelt, in das die Modelle integriert wurden, um ihre praktische Anwendbarkeit zu demonstrieren.Für einen führenden europäischen Stahlhersteller zeigen die Ergebnisse, dass die Einbeziehung des Wärmehaushalts und der Auskleidungsverschleiß der Pfanne sowohl für die operative als auch für die taktische Entscheidungsfindung von zentraler Bedeutung ist, um einen nachhaltigen und kosteneffizienten Betrieb zu erreichen. Bei der Pfannenverteilung wurde ein Kompromiss zwischen direkten (Wiedererwärmung) und indirekten (Ausklei-dungsverbrauch) Emissionen festgestellt. Die optimale Wiedererwärmungsstrategie ist eng mit der wirtschaftlichen Leistung verbunden und bietet die Möglichkeit, die Emissionen um 25 % zu reduzieren und gleichzeitig zusätzliche Einsparungen durch Kohlenstoffabgaben zu erzielen. Bei der Wartungsplanung erwies sich die strategische Reduzierung der täglichen Produktionsrate der Flotte und die Synchronisierung der Pfannenwartung mit Zeiten geringerer Nachfrage als wirksam, um die direkten Emissionen um bis zu 39 % zu reduzieren, indem unnötiges Vorheizen im Vergleich zu konservativen Strategien vermieden wurde. Schließlich zeigte das prototypische Entscheidungshilfetool die Herausforderungen für die praktische Umsetzung auf und hob die Daten- und Modellierungsanforderungen für eine erfolgreiche Einführung hervor.Diese Arbeit macht deutlich, dass zwar eine Optimierung zur Verbesserung der Ressourceneffizienz notwendig ist, jedoch durch die fortgesetzte Abhängigkeit von der Pfannenheizung die Notwendigkeit zusätzlicher Emissionsreduktionsmaßnahmen besteht, insbesondere was den Übergang von Erdgas zu saubereren Brennstoffen betrifft. Daher liefern die Ergebnisse methodische Fortschritte und praktische Erkenntnisse für ein nachhaltiges Management des Pfannenbestands in der Stahlindustrie.Steel is a cornerstone of modern society, yet its production remains one of the most energy and emissions-intensive industrial activities worldwide. Achieving the deep decarbonization targets required by 2050 calls for innovative optimization methods to improve efficiency insteelmaking operations. This thesis investigates how the thermal behavior and refractory degradation of steel ladles, a critical vessel in secondary metallurgy, can be integrated into decision-support frameworks to enhance the sustainability of ladle operations. Towards thisgoal, mixed-integer linear programming and data-driven predictive models were combined to minimize CO2 emissions involved in ladle dispatching and fleet maintenance scheduling to understand optimal operational practices. Finally, a prototype decision-support system was developed, integrating the models to demonstrate their practical applicability.For a leading European steelmaker, the results showed that in corporating the ladle’sthermal balance and refractory degradation is essential for achieving sustainable and costeffective operations in both operational and tactical decision-making. In ladle dispatching,a trade-off between direct (reheating) and indirect (refractory consumption) emissions was identified. The optimal reheating strategy is closely linked to economic performance,with the opportunity to reduce emissions by 25% while unlocking additional savings from carbon levies. In maintenance scheduling, strategically reducing the fleet’s daily production rate and synchronizing ladle maintenance with periods of lower demand proved effective in reducing direct emissions up to 39% by avoiding unnecessary preheating compared to conservative policies. Finally, the prototype decision-support tool demonstrated the challenges for practical implementation, highlighting the data and modeling requirements for successful adoption.This work highlights that while optimization is necessary to improve resource efficiency,the continued reliance on ladle heating underscores the need for additional measures,especially transitioning from natural gas to cleaner fuels. The findings provide methodological advances and practical insights for sustainable ladle fleet management in the steelindustry