324139 research outputs found
Sort by
Professional Service Firms (PSFs) employability skills development for social sciences students in UK higher education
Get PDFBackground: Amid neoliberal marketisation and growing labour market precarity, employability skill development has become central to the UK higher education (HE) agenda. This debate is especially pronounced in the social sciences (SS), which face persistent scrutiny over perceived deficits in employability and graduate work readiness. Employer-reported skills gaps—particularly in Professional Service Firms (PSFs), where effective skill application is critical—have heightened these concerns. Whilst SS graduates are often credited with generic skills, no research has explored how they adapt to PSF careers, or how skills developed in SS programmes align with PSF demands. This study addresses this gap by examining how SS graduates from UK HE acquire employability-related skills and apply them in PSF contexts.Methodology: Sixteen recent SS graduates from UK HE working in PSFs were recruited via convenience and purposive selection. Online semi-structured interviews were conducted and thematically analysed to explore their employability skill development, application, and perceptions of programme preparedness.Findings: UK SS programmes can foster PSF-relevant skills through both curricular and co-curricular ways, though the latter was largely absent in practice. Beyond generic skills, SS education also fosters disciplinary knowledge, with those developed applicable across all stages of PSF work. Whilst skill development varied across degree levels, instructional design, rather than UG or PGT status, emerged as the key determinant. Despite these gains, participants generally viewed SS programmes as offering limited preparation for PSF roles, attributing this to outdated instructional strategies, degree-related stereotypes, restricted opportunities, and a broader negative stance toward SS from the UK government. Notably, external PSF dynamics were observed to hinder the effective application of these skills, warranting further attention.Conclusion: This study offers a holistic understanding of the relationship between SS education and PSF employability through graduate interviews. It addresses a key scholarly gap by analysing the specific nexus between SS education and PSF work, while providing practical insights to guide curriculum reform and strengthen institutional strategies for enhancing graduate employability. The findings challenge the notion that SS is inherently misaligned with PSF careers, emphasising instead that its potential can be fully realised only through deliberate, context-sensitive reform. The study calls for a shift from generic, one-size-fits-all employability agendas toward discipline-sensitive approaches that foster both individual development and systemic change, paving the way for a more integrated and sustainable future for SS education
Secure Extended Care Unit: an exploration of social demographics and clinical characteristics of referred adult mental health patients in Australia: SECU Study
Get PDFBackground and Aims: Secure Extended Care Units (SECU) are low-secure, long-term inpatient rehabilitation for patients with severe mental illnesses. Limited research is available. This study explored the sociodemographic, clinical characteristics and predictors of acceptance in an Australian SECU program over a 5-year period.Methods: A retrospective study design was used to investigate 121 consecutive referrals. The 98 first-time patient referrals were included in the main analysis. Descriptive statistics was used with non-parametric comparisons (Chi-square and Fisher exact test where appropriate). Logistic regression was done to assess the influence of covariates.Results: Most of the Total sample were single males of European ancestry between 25-34 years old with ten years or less of education and receiving disability benefits. Schizophrenia was the predominant diagnosis, with 50% having a personality trait/disorder; substance use was high (82.6 %). More than three-fourths had a history of trauma. Forty-four per cent had a previous forensic admission, with seventy per cent convicted in the past for violence. Physical comorbidity was high (80%), with hepatitis C positivity at 20 per cent. Low service utilisation, like the National Disability Insurance Scheme (NDIS) was noted. Clozapine and Electroconvulsive therapy (ECT) were underutilised. The Median Health of Nations Outcome Scale (HoNOS) was 20 (IQR 14, 23) and the Life Skills Profile (LSP) was 22.5 (IQR 16.25,27). Inpatient setting was the only predictor that influenced acceptance into the program (OR 3.168, 95% CI: 1.129-8.913, p=0.029).Conclusions: Referrals showed a high level of psychosocial-physical complexity, with a range of patient needs, service goals, and high forensic involvement prior to the referral. The study discusses the need for medium and high-secure beds and a new model of care that integrates NDIS and Community Care Units (CCU). A trauma-informed approach that creates holistic treatment plans that include patients and families is indicated
Proteomic evidence for a ginger-flavoured alcoholic beverage found in the corrosion of a 1500–1046 BCE Chinese bronze vessel
No full textDDA data was generated using NanoAcquity-UPLC system (Waters) coupled to a Q-Exactive HF Hybrid Quadrupole-Orbitrap mass spectrometer equipped with an EASY-Spray nano-electrospray ion source (Thermo Fisher Scientific) in August 2024 . Digested peptides were injected into the orbitrap mass spectrometer, a CID fragmentation was promoted. Raw data was generated by Thermo Xcalibur Software 1.05.51.0. PEAKS 8.5 software was used to process the data
Nanopore long-read only genome assembly of clinical Enterobacterales isolates is complete and accurate
Get PDFWhole bacterial genome sequence reconstruction using Oxford Nanopore Technologies (‘Nanopore’) long-read-only sequencing may offer a lower-cost, higher-throughput alternative for pathogen surveillance to ‘hybrid’ assembly with recent improvements in Nanopore sequencing accuracy. We evaluated the accuracy, including plasmid reconstruction, of Nanopore long-read-only genome assemblies of Enterobacterales. We sequenced 92 genomes from clinical Enterobacterales isolates, collected in England under a national surveillance programme, with long-read Nanopore (R10.4.1, Dorado v5.0.0 super-high-accuracy basecalled) and short-read Illumina (NovaSeq) sequencing approaches. Genomes were assembled using three long-read-only (Flye, Hybracter long and Autocycler) and three hybrid assemblers (Hybracter hybrid, Unicycler normal and bold). Three polishing modalities (Medaka v2 with subsampled or un-subsampled long-reads; Polypolish+Pypolca with short-reads) were investigated. Autocycler circularised the most chromosomes [87/92 (95%)]. Plasmid sequence reconstruction was comparable among all assemblers except Flye, all recovering 90–96% of plasmids, although the ‘ground truth’ was uncertain. Flye performed worse than other assemblers on almost all metrics. Autocycler+Medaka (un-subsampled long-reads) was the most accurate long-read-only assembler/polisher combination, comparable to hybrid assemblies [median 0 (IQR: 0–0) single nucleotide variants (SNVs) and 0 (IQR: 0–1) insertions/deletions (indels) per genome; median quality value/Q score 100 (IQR: 64–100)], with only 4/92 genome sequences having >10 SNVs/indels. Medaka polishing with un-subsampled long-reads resulted in small improvements in indels, but not SNVs for both Flye and Autocycler assemblies. Seven-locus multi-locus sequence type, antimicrobial resistance, virulence and stress gene annotation was equivalent across assembler/polisher combinations. Nanopore long-read-only bacterial genome assembly with Autocycler combined with Medaka polishing (using un-subsampled reads) is similarly accurate and possibly more complete than hybrid assemblies, representing a viable alternative for incorporating high-quality genomic data, including plasmids, into Enterobacterales surveillance
Neural networks for learning macroscopic chemotactic sensitivity from microscopic models
No full textThe macroscopic (population-level) dynamics of chemotactic cell movement – arising from underlying microscopic (individual-based) models – are often described by parabolic partial differential equations (PDEs) governing the spatio-temporal evolution of cell concentrations. In certain cases, these macroscopic PDEs can be analytically derived from microscopic models, thereby elucidating the dependence of PDE coefficients on the parameters of the underlying individualbased dynamics. However, such analytical derivations are not always feasible, particularly for more complex or nonlinear microscopic models. In these instances, neural networks offer a promising alternative for estimating the coefficients of macroscopic PDEs directly from data generated by microscopic simulations. In this work, three microscopic models of chemotaxis are investigated. The macroscopic chemotaxis sensitivity is estimated using neural networks, thereby bridging the gap between individual-level behaviours and population-level descriptions. The results are compared with macroscopic PDEs, which can be derived for each model in certain parameter regimes
Multi-task generalization for robotics
Get PDFWhile robots that follow hard-coded instructions have been widely applied in the real world, learning intelligent robots that can autonomously accomplish different tasks in unstructured environments with unforeseen variations remains a key challenge. Inspired by the recent success of foundation models in different domains, robotic learning has been going through a paradigm shift from learning specialist robots on narrow task distributions to learning generalist robots on large-scale multitask data, which enables broader generalization across different dimensions like embodiments, skills and scenarios. In this thesis, we formulate generalist robot learning as a Contextual Markov Decision Process, and investigate two problem settings of generalization across embodiments and skills under this unified framework (Chapter 2). We propose four novel methods to tackle the following three challenges under these two problem settings: model pretraining, inference efficiency and efficient adaptation.In Part II of the thesis, we focus on the first problem setting of cross-embodiment control. In chapter 6, we propose ModuMorph, a Transformer-based universal controller that better models how the optimal policy conditions on the robot morphology via contextual modulation, to improve pretraining. In Chapter 7, we improve inference efficiency of generalist robots via knowledge decoupling, i.e., decoupling the knowledge required to solve different tasks and only activating a compact specialist policy to solve each specific task at test time. We realize knowledge decoupling via the hierarchical architecture of Hypernetworks (HNs), networks that generate the parameters of a base network, and investigate how to successfully train HNs via policy distillation. HyperDistill, our proposed method that combines these two key components, achieves similar performance as ModuMorph, while significantly accelerating inference by two orders of magnitude. In Part III of the thesis, we focus on cross-skill control by following language instructions. In Chapter 8, we extend the knowledge decoupling principle to VisionLanguage-Action (VLA) models, the mainstream approach for language-conditioned control that suffers from high inference cost. We propose HyperVLA to generate a compact policy via an HN that conditions on task context, and investigate several key algorithm design choices to improve the performance of HyperVLA. It achieves similar and even better performance compared to some SOTA VLAs, while significantly improving inference efficiency by two orders of magnitude. Finally in Chapter 9, we investigate how to adapt a pretrained VLA to a new domain with many different tasks in a sample- and computation-efficient way. To achieve these goals, we propose HyperLoRA, which utilizes the strong expressive power of HNs and the parameter-efficient fine-tuning method LoRA to generate task-conditioned LoRA parameters, and significantly outperforms task-agnostic LoRA fine-tuning. In summary, the contributions in this thesis significantly improve model pretraining, inference efficiency and adaptation performance for learning generalist robots. We hope the key ideas developed in this thesis, such as the knowledge decoupling principle and utilizing HNs as a key building block in learning generalist agents, will be utilized more frequently in the future to help build more versatile and efficient foundation models for robotics and other domains
Rescue of extreme hepatectomy mice by primary hepatocyte-derived 3D bio-printed organ transplantation
Get PDFThree-dimensional (3D) bioprinting is an emerging strategy for constructing tissues and organs in vitro. Here, we achieved long-term expansion of primary mouse hepatocytes using a defined medium and constructed liver tissue using 3D bioprinting. The 3D-printed liver tissue demonstrated several essential liver functions and was able to prolong the survival of mice with acute liver failure due to extreme hepatectomy after in vivo transplantation, and the transplanted artificial liver tissue showed distinct functional partitioning. Overall, our results develop a method for long-term in vitro culture of primary hepatocytes and demonstrate the potential of 3D bio-printed liver tissue for clinical translational applications
The Relation between AGN and Host Galaxy Properties in the JWST Era. II. The Merger-driven Evolution of Seyferts at Cosmic Noon
Get PDFIn Paper I, we exploited the unsurpassed resolution and depth of JWST/NIRCam imagery to investigate the relationship between active galactic nuclei (AGN) and host-galaxy properties in the JWST era, finding a correlation between the level of spatial disturbance (as measured by shape asymmetry, AS) and obscuration (NH). Here in Paper II, we report an expansion of our X-ray and infrared analysis of Seyfert-luminosity host galaxies with four additional metrics to the single-metric morphology analysis of Paper I, as well as new samples of inactive control galaxies. This expanded study of one of the largest and most complete, multiwavelength samples of AGN detected at 0.6 < z < 3.8 in the GOODS-South and -North fields, confirms that mergers surprisingly play a significant role in obscured, subquasar AGN host galaxies. Additionally, the pattern of morphological disturbances observed amongst the X-ray- and mid-IR-selected AGN suggests that these represent different phases of AGN evolution tied to a major-merger timeline, as opposed to distinct populations of AGN. These results indicate that mergers are important in triggering subquasar AGN at these redshifts
It’s Just Another Day: Unique Video Captioning by Discriminitive Prompting: It’s Just Another Day: Unique Video Captioning by..
Get PDFLong videos contain many repeating actions, events and shots. These repetitions are frequently given identical captions, which makes it difficult to retrieve the exact desired clip using a text search. In this paper, we formulate the problem of unique captioning: Given multiple clips with the same caption, we generate a new caption for each clip that uniquely identifies it. We propose Captioning by Discriminative Prompting (CDP), which predicts a property that can separate identically captioned clips, and use it to generate unique captions. We introduce two benchmarks for unique captioning, based on egocentric footage and timeloop movies – where repeating actions are common. We demonstrate that captions generated by CDP improve text-to-video R@1 by 15% for egocentric videos and 10% in timeloop movies. https://tobyperrett.github.io/its-just-another-da
Donne’s Architecture of Discernment
Get PDFThis article argues, with extended glances at the poems, that difficulty of view is a vital feature of Donne’s sermons. Difficulties of interpretation are ‘curious frames’ for Donne, ways of attaining a partial knowledge of God, and discerning a ‘better architecture’ beyond. Architectural images in particular are not only valuable means of dividing up and memorizing a sermon’s content, but also provide epistemologically helpful obstructions. Physical sacred spaces also allow Donne to present a vision of Christian life contracted into a small, single area, if the believer attends in the right way, and does not expect to find God too easily. Particular spaces are generalized, made into abstracts for truths which might otherwise seem distant and unassailable. The purpose of the multiple meanings of scripture, and of Donne’s eloquence and deep learning, is not a curious attention to individual details for their own sake. Instead, Donne pursues the listener’s edification, the rebuilding of their life according to God’s plan. The difficulties of building up this Christian life are epitomized in the difficulties of the sermon’s form, but so too are the delights