Acropolis Educational Resources Repository
Not a member yet
31278 research outputs found
Sort by
Lightweight visual question answering using scene graphs
No full textVisual question answering (VQA) is a challenging problem in machine perception, which requires a deep joint understanding of both visual and textual data. Recent research has advanced the automatic generation of high-quality scene graphs from images, while powerful yet elegant models like graph neural networks (GNNs) have shown great power in reasoning over graph-structured data. In this work, we propose to bridge the gap between scene graph generation and VQA by leveraging GNNs. In particular, we design a new model called Conditional Enhanced Graph ATtention network (CE-GAT) to encode pairs of visual and semantic scene graphs with both node and edge features, which is seamlessly integrated with a textual question encoder to generate answers through question-graph conditioning. Moreover, to alleviate the training difficulties of CE-GAT towards VQA, we enforce more useful inductive biases in the scene graphs through novel question-guided graph enriching and pruning. Finally, we evaluate the framework on one of the largest available VQA datasets (namely, GQA) with ground-truth scene graphs, achieving the accuracy of 77.87%, compared with the state of the art (namely, the neural state machine (NSM)), which gives 63.17%. Notably, by leveraging existing scene graphs, our framework is much lighter compared with end-to-end VQA methods (e.g., about 95.3% less parameters than a typical NSM)
Recovering local structure information from high-pressure total scattering experiments
No full textHigh pressure is a powerful thermodynamic tool for exploring the structure and the phase behaviour of the crystalline state, and is now widely used in conventional crystallographic measurements. High-pressure local structure measurements using neutron diffraction have, thus far, been limited by the presence of a strongly scattering, perdeuterated, pressure-transmitting medium (PTM), the signal from which contaminates the resulting pair distribution functions (PDFs). Here, a method is reported for subtracting the pairwise correlations of the commonly used 4:1 methanol:ethanol PTM from neutron PDFs obtained under hydrostatic compression. The method applies a molecular-dynamics-informed empirical correction and a non-negative matrix factorization algorithm to recover the PDF of the pure sample. Proof of principle is demonstrated, producing corrected high-pressure PDFs of simple crystalline materials, Ni and MgO, and benchmarking these against simulated data from the average structure. Finally, the first local structure determination of α-quartz under hydrostatic pressure is presented, extracting compression behaviour of the real-space structure
Archaeomaterials, Innovation, and Technological Change
No full textThe field of archaeomaterials research has enormous potential to shed light on past innovation processes. However, this potential has been only partially recognized outside its immediate practitioners, despite the fact that innovation and technology change are topics of enduring interest in archaeology and the broader social sciences. This review explores the relationship between archaeomaterials research and the interdisciplinary study of innovation, and maps out a path toward greater integration of materials analysis into these discussions.
To foster this integration, this review has three aims. First, I sketch the theoretical landscape of approaches to the study of innovation in archaeology and neighboring disciplines. I trace how theoretical traditions like evolutionary archaeology have influenced archaeomaterials approaches to questions of technological change while also highlighting cases where work by archaeomaterials researchers anticipated trends in the anthropology of technology. Next, I distill a series of core concerns that crosscut these different theoretical perspectives. Finally, I describe examples where archaeomaterials research has deepened scholarly understanding of innovation processes and addressed these core questions. The future of archaeomaterials research lies in engagement with these broader discussions and effective communication of the contributions that materials analysis can make to building a comparative understanding of innovation processes
Inequalities and poverty risks in old age across Europe: The double-edged income effect of pension systems
No full textWhile the sustainability of pension systems facing demographic ageing has been widely discussed, the adequacy of retirement income has often been neglected in current debate. However, considerable poverty and income inequality in old age exists across Europe. Using recent EU‐SILC data (2017/18), the comparative analysis of poverty rates and income inequality in old age shows important cross‐national variations that need to be seen in context of market‐related inequalities but also the specific pension system. Beveridge basic security is not always capable of effectively reducing poverty despite the explicit goal to do so. In addition, private funded pensions may generate social inequality. Some contributory Bismarckian systems are better suited to reduce poverty, but given their focus on status maintenance also reproduce inequality. Poverty rates are low due to encompassing basic pensions in Dutch and some Nordic multipillar systems and in core Central and Eastern European countries. Bismarckian pensions such as in Germany are generating some inequality and medium level of poverty, while France and some Southern European countries perform better on poverty but reproduce larger inequalities. Beveridge systems such as in the United Kingdom and Switzerland with rather meagre basic multipillar systems have relatively medium to high poverty risks. In addition, the Baltic countries and new EU member states in the periphery have the highest poverty rates across Europe. The analysis shows that the minimum income provision of public pension systems matters most for poverty risks, while the overall pension architecture has an impact on reproducing inequality in old age acquired during working life
A school-based social-marketing intervention to promote sexual health in English secondary schools: the Positive Choices pilot cluster RCT
No full textBackground
The UK still has the highest rate of teenage births in western Europe. Teenagers are also the age group most likely to experience unplanned pregnancy, with around half of conceptions in those aged Objectives
To optimise and feasibility-test Positive Choices and then conduct a pilot trial in the south of England assessing whether or not progression to Phase III would be justified in terms of prespecified criteria.
Design
Intervention optimisation and feasibility testing; pilot randomised controlled trial.
Setting
The south of England: optimisation and feasibility-testing in one secondary school; pilot cluster trial in six other secondary schools (four intervention, two control) varying by local deprivation and educational attainment.
Participants
School students in year 8 at baseline, and school staff.
Interventions
Schools were randomised (1 : 2) to control or intervention. The intervention comprised staff training, needs survey, school health promotion council, year 9 curriculum, student-led social marketing, parent information and review of school/local sexual health services.
Main outcome measures
The prespecified criteria for progression to Phase III concerned intervention fidelity of delivery and acceptability; successful randomisation and school retention; survey response rates; and feasible linkage to routine administrative data on pregnancies. The primary health outcome of births was assessed using routine data on births and abortions, and various self-reported secondary sexual health outcomes.
Data sources
The data sources were routine data on births and abortions, baseline and follow-up student surveys, interviews, audio-recordings, observations and logbooks.
Results
The intervention was optimised and feasible in the first secondary school, meeting the fidelity targets other than those for curriculum delivery and criteria for progress to the pilot trial. In the pilot trial, randomisation and school retention were successful. Student response rates in the intervention group and control group were 868 (89.4%) and 298 (84.2%), respectively, at baseline, and 863 (89.0%) and 296 (82.0%), respectively, at follow-up. The target of achieving ≥ 70% fidelity of implementation of essential elements in three schools was achieved. Coverage of relationships and sex education topics was much higher in intervention schools than in control schools. The intervention was acceptable to 80% of students. Interviews with staff indicated strong acceptability. Data linkage was feasible, but there were no exact matches for births or abortions in our cohort. Measures performed well. Poor test–retest reliability on some sexual behaviour measures reflected that this was a cohort of developing adolescents. Qualitative research confirmed the appropriateness of the intervention and theory of change, but suggested some refinements.
Limitations
The optimisation school underwent repeated changes in leadership, which undermined its participation. Moderator analyses were not conducted as these would be very underpowered.
Conclusion
Our findings suggest that this intervention has met prespecified criteria for progression to a Phase III trial.
Future work
Declining prevalence of teenage pregnancy suggests that the primary outcome in a full trial could be replaced by a more comprehensive measure of sexual health. Any future Phase III trial should have a longer lead-in from randomisation to intervention commencement.
Trial registration
Current Controlled Trials ISRCTN12524938.
Funding
This project was funded by the National Institute for Health Research (NIHR) Public Health Research programme and will be published in full in Public Health Research; Vol. 9, No. 1. See the NIHR Journals Library website for further project information
Colocalization analysis of polycystic ovary syndrome to identify potential disease-mediating genes and proteins
No full textPolycystic ovary syndrome (PCOS) is a common complex disease in women with a strong genetic component and downstream consequences for reproductive, metabolic and psychological health. There are currently 19 known PCOS risk loci, primarily identified in women of Han Chinese or European ancestry, and 14 of these risk loci were identified or replicated in a genome-wide association study of PCOS performed in up to 10,074 cases and 103,164 controls of European descent. However, for most of these loci the gene responsible for the association is unknown. We therefore use a Bayesian colocalization approach (Coloc) to highlight genes in PCOS-associated regions that may have a role in mediating the disease risk. We evaluated the posterior probabilities of evidence consistent with shared causal variants between 14 PCOS genetic risk loci and intermediate cellular phenotypes in one protein (N = 3301) and two expression quantitative trait locus datasets (N = 31,684 and N = 80–491). Through these analyses, we identified seven proteins or genes with evidence of a possibly shared causal variant for almost 30% of known PCOS signals, including follicle stimulating hormone and ERBB3, IKZF4, RPS26, SUOX, ZFP36L2, and C8orf49. Several of these potential effector proteins and genes have been implicated in the hypothalamic–pituitary–gonadal signalling pathway and provide an avenue for functional follow-up in order to demonstrate a causal role in PCOS pathophysiology
Evaluation of antigen-detecting and antibody-detecting diagnostic test combinations for diagnosing melioidosis
No full textBackground
Melioidosis, an infectious disease caused by Burkholderia pseudomallei, is endemic in many tropical developing countries and has a high mortality. Here we evaluated combinations of a lateral flow immunoassay (LFI) detecting B. pseudomallei capsular polysaccharide (CPS) and enzyme-linked immunosorbent assays (ELISA) detecting antibodies against hemolysin co-regulated protein (Hcp1) or O-polysaccharide (OPS) for diagnosing melioidosis.
Methodology/Principal findings
We conducted a cohort-based case-control study. Both cases and controls were derived from a prospective observational study of patients presenting with community-acquired infections and sepsis in northeast Thailand (Ubon-sepsis). Cases included 192 patients with a clinical specimen culture positive for B. pseudomallei. Controls included 502 patients who were blood culture positive for Staphylococcus aureus, Escherichia coli or Klebsiella pneumoniae or were polymerase chain reaction assay positive for malaria or dengue. Serum samples collected within 24 hours of admission were stored and tested using a CPS-LFI, Hcp1-ELISA and OPS-ELISA. When assessing diagnostic tests in combination, results were considered positive if either test was positive. We selected ELISA cut-offs corresponding to a specificity of 95%. Using a positive cut-off OD of 2.912 for Hcp1-ELISA, the combination of the CPS-LFI and Hcp1-ELISA had a sensitivity of 67.7% (130/192 case patients) and a specificity of 95.0% (477/502 control patients). The sensitivity of the combination (67.7%) was higher than that of the CPS-LFI alone (31.3%, p<0.001) and that of Hcp1-ELISA alone (53.6%, p<0.001). A similar phenomenon was also observed for the combination of CPS-LFI and OPS-ELISA. In case patients, positivity of the CPS-LFI was associated with a short duration of symptoms, high modified Sequential (sepsis-related) Organ Failure Assessment (SOFA) score, bacteraemia and mortality outcome, while positivity of Hcp1-ELISA was associated with a longer duration of symptoms, low modified SOFA score, non-bacteraemia and survival outcome.
Conclusions/Significance
A combination of antigen-antibody diagnostic tests increased the sensitivity of melioidosis diagnosis over individual tests while preserving high specificity. Point-of-care tests for melioidosis based on the use of combination assays should be further developed and evaluated
Historical national accounting and dating the Great Divergence
No full textBy offering a particular interpretation of the new evidence on historical national accounting, Goldstone argues for a return to the Pomeranz (2000) version of the Great Divergence, beginning only after 1800. However, he fails to distinguish between two very different patterns of pre-industrial growth: (1) alternating episodes of growing and shrinking without any long-term trend in per capita income and (2) episodes of growing interspersed by per capita incomes remaining on a plateau, so that per capita GDP trends upwards over the long run. The latter dynamic pattern occurred in Britain and Holland from the mid-fourteenth century, so that Northwest Europe first edged ahead of the Yangzi delta region of China in the eighteenth century
Possible involvement of the nutrient and energy sensors mTORC1 and AMPK in cell fate diversification in a non-metazoan organism
No full textmTORC1 and AMPK are mutually antagonistic sensors of nutrient and energy status that have been implicated in many human diseases including cancer, Alzheimer’s disease, obesity and type 2 diabetes. Starved cells of the social amoeba Dictyostelium discoideum aggregate and eventually form fruiting bodies consisting of stalk cells and spores. We focus on how this bifurcation of cell fate is achieved. During growth mTORC1 is highly active and AMPK relatively inactive. Upon starvation, AMPK is activated and mTORC1 inhibited; cell division is arrested and autophagy induced. After aggregation, a minority of the cells (prestalk cells) continue to express much the same set of developmental genes as during aggregation, but the majority (prespore cells) switch to the prespore program. We describe evidence suggesting that overexpressing AMPK increases the proportion of prestalk cells, as does inhibiting mTORC1. Furthermore, stimulating the acidification of intracellular acidic compartments likewise increases the proportion of prestalk cells, while inhibiting acidification favors the spore pathway. We conclude that the choice between the prestalk and the prespore pathways of cell differentiation may depend on the relative strength of the activities of AMPK and mTORC1, and that these may be controlled by the acidity of intracellular acidic compartments/lysosomes (pHv), cells with low pHv compartments having high AMPK activity/low mTORC1 activity, and those with high pHv compartments having high mTORC1/low AMPK activity. Increased insight into the regulation and downstream consequences of this switch should increase our understanding of its potential role in human diseases, and indicate possible therapeutic interventions
OpenIFS@home version 1: a citizen science project for ensembleweather and climate forecasting
No full textWeather forecasts rely heavily on general circulation models of the atmosphere and other components of the Earth system. National meteorological and hydrological services and intergovernmental organisations, such as the European Centre for Medium-Range Weather Forecasts (ECMWF), provide routine operational forecasts on a range of spatio-temporal scales, by running these models in high resolution on state-of-the-art high-performance computing systems. Such operational forecasts are very demanding in terms of computing resources. To facilitate the use of a weather forecast model for research and training purposes outside the operational environment, ECMWF provides a portable version of its numerical weather forecast model, OpenIFS, for use by universities and other research institutes on their own computing systems.
In this paper, we describe a new project (OpenIFS@home) that combines OpenIFS with a citizen science approach to involve the general public in helping conduct scientific experiments. Volunteers from across the world can run OpenIFS@home on their computers at home and the results of these simulations can be combined into large forecast ensembles. The infrastructure of such distributed computing experiments is based on our experience and expertise with the climateprediction.net and weather@home systems.
In order to validate this first use of OpenIFS in a volunteer computing framework, we present results from ensembles of forecast simulations of tropical cyclone Karl from September 2016, studied during the NAWDEX field campaign. This cyclone underwent extratropical transition and intensified in mid-latitudes to give rise to an intense jet-streak near Scotland and heavy rainfall over Norway. For the validation we use a two thousand member ensemble of OpenIFS run on the OpenIFS@home volunteer framework and a smaller ensemble of the size of operational forecasts using ECMWF’s forecast model in 2016 run on the ECMWF supercomputer with the same horizontal resolution as OpenIFS@home. We present ensemble statistics that illustrate the reliability and accuracy of the OpenIFS@home forecasts as well as discussing the use of large ensembles in the context of forecasting extreme events