Articles tagged with Data Sets
A new study from Stanford University's Immigration Policy Lab found that many refugees in the US move to different states soon after arrival, primarily seeking better job markets and social networks. Refugees are more likely to leave high-unemployment states and join those with booming economies.
Experts have developed a platform for self-testing AI medical services, allowing for automated validation and improvement. The platform provides an opportunity to fine-tune algorithms with unlimited access to data instances, minimizing human factor manipulation.
Researchers at Texas A&M University are using advanced machine-learning analysis to reveal the evolution of cracks in rock and concrete. By combining data from multiple sources, including sound waves, electromagnetics, and pressure measurements, they aim to improve our understanding of crack damage and development.
Researchers developed a computational tool called PolyA-miner to analyze alternative polyadenylation (APA) sites in RNA strands. The tool precisely identifies novel APA sites that were not detected by traditional analytical approaches, revealing new insights into gene regulation.
Tong and collaborators are developing a model-ready emission dataset for CAM-CMAQ's wildfire forecasting system, aiming to improve accuracy and mitigate adverse fire effects. The dataset will also incorporate aerosol attenuation of photolysis in smoke plumes.
A team of researchers has quantified human activity across Antarctica, revealing that almost every area has been visited. Biodiversity is not well-represented within these areas, highlighting the need for swift action to declare new protected areas.
A major loss of grassland has occurred in Great Britain since 1990, with a net reduction of 7,668 km2 or 1.9 million acres, as woodland area increased by 5,236 km2 and urban areas expanded by 3,376 km2 between 1990 and 2015. The majority of this loss and increase occurred in Scotland and England respectively.
Researchers from Caltech and institutions like Northwestern University used deep learning and supercomputing to identify Nyx, a product of a long-ago galaxy merger. The discovery provides the first indication that a dwarf galaxy merged with the Milky Way disk.
Researchers at Cornell University used AI to investigate how reflection changes images, discovering clues like facial features and beards that can differentiate originals from reflections. The study has implications for training machine learning models and detecting faked images.
Researchers developed a new set of computational tools to identify cell-type specific methylation patterns, known as molecular barcodes, in complex cell mixtures. These new methods can distinguish between different cell types in tissues made up of multiple cell types.
A large-scale database of ground reaction force data from over 2,000 patients has been made publicly available for research purposes. The GAITREC database includes information on patients with joint transplants, fractures, and ligament injuries, providing a valuable resource for clinicians and researchers.
A team of researchers at Carnegie Mellon University has developed an automated method for making communications more polite, using a large dataset of labeled sentences. The system can restructure nonpolite directives or add words to make them more well-mannered, while maintaining the original meaning.
A team has devised deep-learning and other computational approaches that dramatically reduce image-analysis time by orders of magnitude. They report their results in Nature Biotechnology, accelerating image analysis in three major ways: deconvolution, 3D registration, and complex deconvolution.
By extending available data, researchers developed a method to estimate mass concentrations of particulate matter from humidity and visibility measurements. This approach has the potential to provide broader understanding of how particulate matter evolves and improve visibility in daily life.
Researchers have created a new map revealing the Earth's mantle at a depth of 3,000 kilometers by analyzing thousands of seismic waves collected over 30 years. The map shows hot and dense regions below Hawaii and French Polynesia.
Researchers at King Abdullah University of Science & Technology devised a new analytical tool to predict flood risk by adapting a classical statistical model for analyzing extreme rainfall in large datasets. The model demonstrated potential in capturing observed patterns in northeast America, promising improved prediction capabilities.
A multivariate analysis of sea level data reveals that barystatic and steric contributions drive most interannual variability in GMSL, with ENSO playing a significant role. The study highlights the magnitude of these contributions associated with ENSO events.
Using air conditioners increases electricity bills on average by 42%, particularly affecting poorer households. Climate change amplifies this impact, especially in Europe and North America.
A new methodology for assessing the credibility of historical global land use/cover datasets has been proposed, addressing temporal and spatial changes. The approach evaluates accuracy, rationality, and likelihood assessments through five case studies, providing a framework for improving data quality.
The gnomAD Consortium has published its first major studies of human genetic variation, revealing new insights into rare types of genetic variation. The research provides better tools for clinical geneticists to diagnose patients with rare genetic diseases and evaluate proposed drug targets.
A study found significant gender imbalance in medical imaging datasets used to train classifiers, leading to biased AI-based diagnostic systems. The analysis showed that underrepresented groups were misclassified more frequently than the majority group.
A Tel Aviv University-led study published in Nature found that complexity in analytical methods contributes to variability in research outcomes. Researchers analyzed the same dataset using different analysis methods, resulting in varying conclusions. The study highlights the need for improved methodology and data sharing to advance sci...
A new soil moisture product has been developed for mainland China, utilizing a dual-pass data assimilation system that auto-calibrates model parameters. The product provides gridded soil moisture data with a spatial resolution of 0.25° over China, showing consistent patterns with precipitation and evaporation.
The Pofatu Database provides a comprehensive compilation of geochemical analyses and contextual information for archaeological sources and artefacts. It enables researchers to reconstruct ancient strategies of raw material and artefact procurement, facilitating comparability and reproducibility in provenance studies.
Researchers developed a deep-learning framework called Morpheus to perform pixel-level morphological classifications of objects in astronomical images. The tool can handle complex images and provide detailed classification results with confidence levels.
Researchers found that winning back lost customers can lead to increased profits, but requires a failure-tolerant organizational culture that encourages open discussion and accountability. Successful reacquisition management also involves establishing guidelines for employees to follow when addressing customer defections.
A new high-resolution 3D map of the mouse brain has been published, providing a reference atlas for the neuroscience community. The map enables whole-brain studies and improves research by allowing researchers to precisely co-register different types of data, enabling bigger-picture views and comparisons.
The TACC COVID-19 Twitter dataset enables researchers to analyze social media communications and identify trends in pandemic responses. The dataset, which contains over 40 million tweets, can be used for topic modeling, entity analysis, and event detection, facilitating discoveries about the spread of misinformation and racist messaging.
A new study found that men are over twice as likely to die from COVID-19, with older men and those with underlying conditions being particularly at risk. The research suggests that additional supportive care may be necessary for these groups.
Researchers propose a new methodology to alter survey datasets to protect consumer privacy, preserving reasonable accuracy. Survey data is vulnerable to breaches and unauthorized use, with some employees stealing sensitive data from their former companies.
A new study finds that teachers exhibit almost identical levels of pro-White racial bias as non-teachers, highlighting the need for additional support and training to mitigate implicit biases in schools. The research suggests that schools are microcosms of society, and teachers require help in combating their biases.
The new dataset, published in PLOS ONE, analyzed 15,506 instances of social scientists' testimony and found that economists testified more than four times as often as political scientists and over 10 times as often as sociologists. Economists also represented think tanks, while anthropologists had the lowest rate of testimony.
Researchers found that state-of-the-art ASR systems performed worse on black speakers than white speakers, with error rates of 0.35 and 0.19 words per hour respectively. The study attributes these disparities to limitations in the acoustic models' ability to capture African American Vernacular English pronunciation and prosody.
Researchers at University of Münster develop AI tool to predict reaction outcomes using molecular structures, enabling accurate predictions for yields and stereoselectivities. The model can be applied to diverse reactions and is expected to significantly change the approach to chemical syntheses.
Using DES data, researchers found more than 300 trans-Neptunian objects, including 245 discoveries made by DES. The method developed by Pedro Bernardinelli can also be used to search for TNOs in upcoming astronomy surveys.
Agriculture-Vision dataset enables farmers to analyze aerial images and gain actionable insights into crop performance. The dataset, developed by researchers at the University of Illinois Grainger College of Engineering and Intelinair, includes over 100,000 images from corn and soybean fields across the Midwest.
A team of researchers developed a standardized survey tool to collect data on rural households. The Rural Household Multi-Indicator Survey (RHoMIS) includes over 30,000 interviews from 33 countries and provides insights into smallholder farming practices, climate change, and social inclusion.
Researchers found that social banks produce smaller returns to their owners, remunerate depositors at below market interest rates, and grant loans below market interest rates. This supports the hypothesis that ownership explains the results, as stakeholder banks exhibit different interest rate behavior.
The World Karst Spring hydrograph dataset (WoKaS) offers a comprehensive collection of over 400 karst spring discharge data, providing valuable insights into the world's fastest-flowing groundwater. This database supports trend analyses, impact studies, and model evaluations for sustainable water management.
A system created by MIT researchers can pinpoint and replace specific information in Wikipedia sentences while retaining humanlike grammar and style. This technology has potential to automatically update factual inconsistencies in Wikipedia articles, reducing time and effort spent by human editors.
A study of over 760,000 restaurant reviews found that location bias affects 98% of restaurants, altering customer consideration by up to 16%. Improved ranking systems can lead to higher reviewer satisfaction (up to 12%) and more diversified top-restaurant recommendations (up to 24%).
A new AI method combines incomplete medical datasets to spot causal relationships, allowing researchers to analyze expensive or difficult-to-run trials. The algorithm uses quantum cryptography-inspired principles to determine the most disordered variable as the cause.
A research project compares records from Augsburg and Aberdeen to analyze city administrations, evolution, similarities, and differences. The collaboration aims to develop new analysis techniques and share ideas on challenges and solutions.
A new interactive map from Harvard University reveals nonviolent uprisings are more successful in achieving their goals, contrary to long-held assumptions. The dataset covers 1945-2014 and includes hundreds of maximalist campaigns globally.
Researchers developed RoadTagger to predict lane counts and road types with high accuracy, helping improve GPS navigation in unfamiliar locations. The model uses a combination of neural network architectures to automatically tag road features, enabling the creation of accurate digital maps.
Researchers develop technique to evaluate reliability of patient risk models, which often fail to accurately predict outcomes. The method generates an unreliability score, indicating when a model's predictions are less trustworthy, helping doctors avoid ineffective or unnecessarily risky treatments.
The Ruhr-University Bochum team used a combination of data analytics and physical models to predict material properties, winning second place in an international machine learning competition. Their algorithm was published in NPJ Computational Materials and has been shown to be transferable to different material classes.
Researchers found that standard pathology tests provide better information on likely outcome and best treatment options for patients with bladder cancer than molecular subtyping techniques. The study suggests that more study is needed before molecular subtypes are used to guide patient care.
Researchers from the Chinese Academy of Sciences conducted model perturbation experiments to improve monsoon predictions. The study sheds light on the complex atmosphere-ocean-land interactions driving monsoon systems and their regional variations.
Researchers at Dartmouth College and Harvard Medical School have confirmed that single-cell eukaryote Stentor roeseli can make complex decisions, demonstrating avoidance behavior and a hierarchy of responses to stimulus. The study uses modern technology to recreate an early experiment that was debunked decades ago.
A new analysis of a dataset of over 70,000 North American migratory birds found that body size decreased in all 52 species, with statistically significant declines in 49 species, while wing length increased in 40 species. The study suggests that warming temperatures may be causing these changes as an adaptation to climate change.
A team from Colorado State University has created a novel dataset, AquaSat, by merging large public datasets of water quality observations with satellite imagery. This 'symphony of data' provides over 6 million water quality observations, unlocking powerful new applications in remote sensing of water quality. The study aims to improve ...
Researchers developed AI models for chest X-ray interpretation that can detect fractures, nodules, opacity and pneumothorax as effectively as experienced radiologists. The models were trained on large datasets and evaluated using a panel of radiologists to increase expert consensus and accuracy.
A recent study by researchers at SESYNC found no concrete evidence to attribute the effects of the Justinianic Plague to the pandemic. The team examined diverse datasets, including written sources, inscriptions, and pollen samples, but found that trends continued without change before and after the plague outbreak.
Researchers can now visualize data from samples containing tens of millions of cells with unprecedented resolution, pinpointing previously undetectable features that distinguish diseased samples from controls. This advancement may lead to discovery of novel cell types to therapeutically target diseases.
A new method accelerates template creation for medical-image analysis, generating brain scan templates based on patient attributes such as age and sex. The model can synthesize atlases from sparse data, improving disease diagnosis accuracy.
A new program, CRUP, simplifies analysis of enhancer regions and their target genes in the genome. The tool identifies active enhancers and links them to their respective genes, helping researchers understand complex diseases.
Researchers found that satellite data sets are reliable substitutes for reproducing water vapor features over the Tibetan Plateau, while reanalysis data sets have varying quality depending on time scales considered. A skill-weighted ensemble mean of reanalysis data shows promise for future studies.
A new machine learning method called SNIPER has successfully applied to nine cell lines, providing missing information for how DNA is organized in the cell nucleus. This enables scientists to study differences in spatial organization related to subcompartments across those lines.
Researchers at MIT have compiled a comprehensive dataset of robotic pushing dynamics, capturing the behavior of hundreds of different objects. This dataset, called Omnipush, enables robots to learn fundamental object manipulation tasks, such as reorienting and inspecting objects, by training on a diverse range of pushing scenarios.