13463 research outputs found
Sort by
Understanding Foundation Models: Are We Back in 1924?
Get PDFThis position paper explores the rapid development of Foundation Models (FMs) in AI and their implications for intelligence and reasoning. It examines the characteristics of FMs, including their training on vast datasets and use of embedding spaces to capture semantic relationships. The paper discusses recent advancements in FMs’ reasoning abilities which we argue cannot be attributed to increased model size but to
novel training techniques which yield learning phenomena like
grokking. It also addresses the challenges in benchmarking FMs
and compares their structure to the human brain. We argue
that while FMs show promising developments in reasoning and
knowledge representation, understanding their inner workings
remains a significant challenge, similar to ongoing efforts in
neuroscience to comprehend human brain function. Despite
having some similarities, fundamental differences between FMs
and the structure of human brain warn us against making direct
comparisons or expecting neuroscience to provide immediate
insights into FM function
The EU Micro-interventionism in Kosovo
Get PDFThis article critically examines the European Union's (EU) micro-interventionism in Kosovo, focusing on its implications for statebuilding and institutional reforms. It argues that, while the EU's micro-interventions aim to strengthen governance and promote European integration, they often undermine local ownership, provoke resistance, and yield superficial progress. The study analyses three key domains: rule of law reforms, where embedded EU officials and subcontracted experts have led to excessive micromanagement and reduced local agency; public administration reforms, characterised by externally imposed, short-term fixes rather than sustainable improvements; and micro-diplomacy in Kosovo-Serbia relations, which has fostered arbitrary negotiation practices and limited long-term conflict resolution. The article highlights the paradox of EU interventionism - intended to promote stability yet often resulting in governance inefficiencies and heightened local sovereignty demands. By providing a micro-political analysis, it contributes to debates on international statebuilding, emphasizing the need for more collaborative and locally driven approaches
Agreements on Normalisation and Statehood
No full textThis chapter explores the complex and multifaceted nature of agreements on normalisation and statehood, which play a critical role in resolving conflicts and modifying sovereign relations between states. It begins by examining how such agreements aim to create, restore, or maintain normal relations, addressing key issues like sovereignty, territorial integrity, and diplomatic recognition. The chapter categorises these agreements into two main groups: those that facilitate state creation through secession and those that focus on normalising diplomatic relations between existing states. It highlights the importance of these agreements in managing self-determination conflicts, which are often sensitive and intractable. The chapter also delves into the political and legal controversies that arise in such agreements, particularly their use of vague language and the challenges posed by differing legal interpretations. Finally, it examines the involvement of third-party mediators and the complex dynamics of enforcing and sustaining these agreements, given their significant political and legal implications
High-Resolution FBG Strain Sensing With DualComb Interrogation and Optimized Signal Processing
Get PDFIn this work, we present a high-resolution strain sensing system based on Dual Optical Frequency Comb (DOFC) interrogation of Fiber Bragg Gratings (FBGs), employing optimized signal processing methods for both reflection and transmission spectra. Femtosecond (fs) written FBGs with nonuniform profiles with bandwidths of ∼0.5 nm were interrogated
using a mutually coherent DOFC generated by externally injected
gain-switched lasers (EI-GSL). Different algorithms were used to
process the DOFC signal to extract the FBG frequency shifts
caused by the various strain increments. Using these approaches,
we could detect strains with a sub-µε resolution (0.32 µε),
sensitivities ranging from 0.7 - 1 pm/µε, and a dynamic range of
422 µε. These methods increased low-strain sensitivity typical of
DOFC interrogation systems, enabling precise monitoring of
standard off-the-shelf fs-written FBGs. The DOFC system showed
enhanced linearity (R2 ∼0.98) and Figure of Merit (FoM - 2.58/µε)
at lower strains than a standard commercial interrogator (R2
∼0.82 and FoM – 0.88/µε). Our findings highlight the potential of
the DOFC-FBG interrogation as a powerful tool for real-time
high-resolution sensing applications, ranging from structural
health monitoring to biomedical diagnostics
Assessing the benefits of digital game-based learning with Minecraft in children, adolescents and young adults: A broad systematic review
Get PDFMinecraft is a popular digital game-based learning (DGBL) tool used in learning environments around the world. With increasing emphasis on evidenceinformed practice and policymaking in educational systems, this paper systematically reviews the evidence base behind the use of Minecraft in improving academic, cognitive, motivational-affective and
social outcomes. We searched seven databases and other secondary sources up until February 2024. Pre-post intervention studies that evaluated the use of Minecraft (randomised, non-randomised
and single-group studies) in children, adolescents and young adults were included. Twenty-nine studies satisfied our inclusion criteria. Studies primarily investigated Minecraft as a tool to improve cognitive (n=7) and academic (n=16) outcomes. Studies reported positive effects on spatial thinking (n=2), creativity (n=3), critical thinking (n=1), mathematics
(n=5), science (n=4) and language (n=2). However, all studies included in this review had a medium or high risk of bias. Therefore, while Minecraft demonstrates good potential in improving these outcomes,
effects need to be replicated in more rigorous studies. Suggestions for future research are discussed. This review provides timely evidence for educators, school leaders and policymakers regarding the use
of DGBL with Minecraft
Optimizing solvent systems for electrospun PLGA scaffolds: effects on microstructure and mechanical properties for biomedical applications
Get PDFElectrospun scaffolds fabricated from poly(lactic-co-glycolic acid) (PLGA) have garnered widespread interest in biomedical applications due to their ability to mimic the extracellular matrix (ECM) structure
with a tunable degradability profile. The properties of electrospun scaffolds are meticulously tailored for specific applications through the adjustment of polymer properties, solution parameters, and processing
conditions. Solvent selection is crucial, influencing polymer spinnability and scaffold topographical, physical and mechanical features. Hansen solubility theory aids in predicting suitable solvent systems. The absence of specific data prompted a solubility experiment to determine Hansen solubility parameters for PLGA. Subsequently, various solvent systems were investigated for their impact on the microstructure of electrospun PLGA scaffolds. Optimizing the electrospinning process resulted in fibrous scaffolds with consistent average fibre diameter from different solvent systems, allowing a focused examination of the solvent's isolated influence on mechanical properties. PLGA samples electrospun
using hexafluoro isopropanol (HFIP) displayed lower Young's modulus and ultimate tensile strength but higher failure strains than those created using binary solvent systems composed of tetrahydrofuran
(THF), dichloromethane (DCM), and dimethylformamide (DMF). This research advances the understanding and optimization of electrospun PLGA scaffolds, enhancing their potential for biomedical applications
Equity and Infrastructure Asset Management: Promoting Community Outcomes in the Water Sector
Get PDFWater utilities across the United States are being tasked with paying increased attention to inequities experienced in the communities they serve. This paper examines whether a widely used decision process within the sector, infrastructure asset management, is compatible with this desire to advance equity. Using three dimensions of equity - distributional, procedural, and recognitional - to interrogate asset management as currently practiced in the US, we question some of its deeply held principles. The Environmental Protection Agency has five core asset management questions which focus on asset systems, sustainability, and long-term costs to the utility; we propose that different core questions are needed to reconcile the asset management system with equity and realize larger community outcomes. A case study is presented showing one utility’s attempt to do that by basing their asset management system in the needs of their oppressed communities. In the end, the authors of this paper remain skeptical about using asset management to drive community equity. Asset management was invented and evolved in a larger system that produces racial inequities, and we are not confident that it can be transformed enough to drive equity
Sequence-Dependent Shape and Stiffness of DNA and RNA Double Helices: Hexanucleotide Scale and Beyond
Get PDFThe structure and deformability of double-stranded DNA and RNA depend on the sequence of bases, affecting biological processes and nanostructure design, but this dependence is incompletely understood. Here we present mechanical properties of DNA and RNA duplexes inferred from atomic-resolution, explicit-solvent molecular dynamics (MD) simulations of 107 DNA and 107 RNA oligomers containing all hexanucleotide sequences. In addition to the level of rigid bases, minor and major grooves, we probe the length and sequence dependence of global material constants such as persistence lengths, stretching and twisting rigidities. We propose a simple model to predict sequencedependent shape and nonlocal, harmonic stiffness for an arbitrary sequence, validate it on an independent set of MD simulations for DNA and RNA duplexes containing all pentamers, and demonstrate its utility in various applications. The large amount of the simulated data enabled us to study rare events, such as base-pair opening, or flips of the A-RNA sugar pucker into the B domain and the related dynamics of the 2′-OH group. Together, this work provides a comprehensive sequence-specific description of DNA and RNA duplex mechanics, forming a baseline for further research and allowing for a broad range of applications
The Use of Computational Approaches to Design Nanodelivery Systems
Get PDFNano-based drug delivery systems present a promising approach to improve the efficacy and safety of therapeutics by enabling targeted drug transport and controlled release. In parallel, computational approaches—particularly Molecular Dynamics (MD) simulations and Artificial Intelligence (AI)—have emerged as transformative tools to accelerate nanocarrier design and optimise their properties. MD simulations provide atomic-to-mesoscale insights into nanoparticle interactions with biological membranes, elucidating how factors such as surface charge density, ligand functionalisation and nanoparticle size affect cellular
uptake and stability. Complementing MD simulations, AI-driven models accelerate the discovery of lipid-based nanoparticle formulations by analysing vast chemical datasets and predicting optimal structures for gene delivery and vaccine development. By harnessing these computational approaches, researchers can rapidly refine nanoparticle composition to improve biocompatibility, reduce toxicity and achieve more precise drug targeting. This review synthesises key advances in MD simulations and AI for two leading nanoparticle platforms (gold and lipid nanoparticles) and highlights their role in enhancing therapeutic
performance. We evaluate how in silico models guide experimental validation, inform rational design strategies and ultimately streamline the transition from bench to bedside. Finally, we address key challenges such as data scarcity and complex in vivo dynamics and propose future directions for integrating computational insights into next generation
nanodelivery systems
Plain Sailing or Choppy Water? Maintaining Interpersonal Trusting Relationships in Times of Uncertainty
Get PDFInterpersonal trusting relationships frequently experience relational threats that require both parties to engage actively in trust maintenance efforts. Yet, trust research has tended to focus on trust formation, or trust repair in the case of a violation, and offers us little insight regarding
how these more ambiguous threats to trusting relationships are experienced and overcome relationally. To provide novel insight on this topic, this exploratory study gathers dyadic interview data from 26 manager–employee trusting relationships regarding their experience of relational threats and their proactive efforts to overcome these negative relational experiences. Findings show that the experience of a relational threat triggers a three-stage trust maintenance process
that includes an assessment phase, an active maintenance phase, and an outcome phase. Threats are assessed at the individual level via cognitive and affective sensemaking, while trust maintenance efforts (creating a shared mental model, cognitive and structural reassurance, and dyadic problem solving) require dyadic counterparts to act with mutual agency to overcome the relational threat and avoid a loss of trust. Trust maintenance processes support dyads to either maintain or strengthen their existing trusting relationships. Our findings advance our theoretical understanding of interpersonal trust maintenance by demonstrating that this process unfolds across three phases and can lead to different outcomes for dyads’ trusting relationships. We offer
practical guidelines to safeguard existing trusting relationships, as well as a new agenda for trust scholars to extend our theorizing