Articles tagged with Computer Modeling
Researchers developed a framework called SigLLM that uses large language models to detect anomalies in time-series data. The approach converts time-series data into text-based inputs and can be deployed right out of the box, offering an efficient anomaly detection solution for complex systems.
The UFZ forest condition monitor tracks dynamic changes in Germany's forests due to climate change extreme events. The index displays seasonal and annual maps of forest condition, highlighting areas with severe damage, such as the Harz region.
Researchers at University of Bath and Technical University of Darmstadt found that large language models like ChatGPT cannot learn independently or acquire new skills, making them controllable and predictable. The study concluded that LLMs remain inherently safe, but misuse is still possible.
Researchers at Carnegie Mellon University propose guidelines for using interpretable machine learning methods in computational biology to tackle complex problems. The guidelines address pitfalls such as relying on a single method and cherry-picking results, emphasizing the need for multiple approaches and human-centric considerations.
A new interpretable deep learning modeling architecture, LACG, is proposed to handle complex variable interactions in chemical processes. It achieves high-accuracy modeling results via the coupling of three sub-modules based on underlying logics. The model outperforms widely used neural networks in terms of interpretability.
Researchers found that powerful waves triggered deep within the Earth can cause continental surfaces to rise by over a kilometre. The study explains why parts of continents experience substantial uplift and erosion, forming sweeping elevated regions known as plateaus.
Researchers developed a novel clustering technique that considers both basic characteristics and target material properties, enabling the categorization of over 1,000 oxides into material groups. This approach uses machine learning to predict target properties and incorporates basic feature information into the analysis.
A research team at Iowa State University has developed artificial intelligence technology that can model and understand complex chemical reactions, including those involved in ammonia production. The technology uses reinforcement learning to identify the optimal reaction pathway, promising to reduce production costs and emissions.
A new study from the University of Illinois Chicago proposes an alternative theory for the formation of Earth's continents, challenging the long-held leading theory. The researchers used computer models to investigate the origin of Archaean zircons, which date back to 2.5-4 billion years ago.
Researchers developed a computer model to examine the dynamics underlying suicide contagion after the suicides of Robin Williams in 2014 and Kate Spade and Anthony Bourdain in 2018. The findings provide a framework for quantifying suicidal contagion, allowing for better understanding, prevention, and containment of its spread.
Researchers have introduced a new AI calibration method called Thermometer, which enables efficient calibration of large language models for various tasks. This technique leverages temperature scaling to adjust a model's confidence and can generalize to new tasks without requiring additional labeled data.
Researchers developed a machine-learning framework that can predict phonon dispersion relations up to 1,000 times faster than other AI-based techniques, with comparable or even better accuracy. This method could help engineers design more efficient power generation systems and develop faster microelectronics.
Researchers investigated the effects of initial microbiota on microbial succession during eggplant fermentation, finding two distinct patterns of LAB dominance. They also identified Lactiplantibacillus plantarum as a primary contributor to lactic acid production and nutrient content.
A new method called D-REC uses a digital twin to predict which data users will need, improving edge caching decisions. The digital twin takes real-time data from the wireless network and conducts simulations to make accurate predictions.
A model simulates the group motion of fish based on visual cues, showing that each fish selects a single target and traces its motion. The model reproduces various collective patterns, including rotating vortexes and turning motions.
A new AI model can identify certain stages of ductal carcinoma in situ (DCIS), a type of pre-invasive breast cancer, that are likely to progress to invasive cancer. The model uses imaging and machine learning to analyze tissue samples and determine the stage of DCIS based on cell arrangement and organization.
The Special Report explores expert perspectives on deploying AI in radiology, emphasizing the need for trust, reproducibility, explainability, and accountability. Radiologists must be able to trust in AI systems' design and receive adequate training, while establishing clear guidelines regarding clinical accountability.
A new study proposes a predictive home energy management system with a customizable bidirectional real-time pricing mechanism to promote residential demand response and reduce peak loads. The system enhances user comfort and accuracy of forecasting, while also providing cost savings.
A new video-processing technique developed at the University of Florida uses artificial intelligence to track Parkinson's disease progression in patients, allowing neurologists to better monitor their condition and quality of life. The system analyzes video recordings of patients performing the finger-tapping test, a standard test for ...
Researchers found that large language models perform poorly in high-stakes situations despite being better than smaller models, due to misalignment with human generalization function. Human generalization, which involves forming beliefs about others' abilities, plays a significant role in LLM performance and deployment.
Researchers at USC developed a new method to accurately predict wildfire spread using satellite data and artificial intelligence. The model offers a potential breakthrough in wildfire management and emergency response, providing more precise and timely data for firefighters and evacuation teams battling wildfires.
A study found that large language models, despite accuracy in medical exams, fail to consistently request necessary examinations and often deviate from treatment guidelines. In comparison to human doctors, AI diagnoses achieved lower accuracy rates, highlighting concerns about their suitability for everyday clinical practice.
UCF researchers George Atia and Yue Wang received a $1.2 million DARPA grant to develop AI-based technologies that can help autonomous systems adapt to unknown variables and overcome simulation-to-real gap issues.
Researchers at Georgia Tech developed a neural network that makes decisions similar to humans, using Bayesian neural networks and evidence accumulation processes. The model outperformed rival deterministic models and showed improved performance in high-speed scenarios, demonstrating its potential for more accurate decision-making.
A recent study predicts that lakes worldwide will experience unprecedented surface and subsurface warming, leading to severe disruptions in ecosystems. Tropical lakes are expected to be the first to emerge from natural temperature bounds, while high-latitude lakes may shield their subsurface layers from surface warming.
Researchers developed an AI model to analyze heart MRI scans, which can save NHS time and resources. The model provides a complete analysis of the entire heart using a view that shows all four chambers, leading to faster and more accurate diagnosis of heart failure and other cardiac conditions.
Researchers at the University of California - Riverside have proposed a chain of quantum magnetic objects called spin centers that can simulate exotic magnetic phases of matter. This breakthrough could lead to more efficient ways of storing and transferring information, as well as the development of room temperature quantum computers.
GenSQL integrates a tabular dataset and a generative probabilistic AI model to analyze complex tabular data. It can detect anomalies, predict outcomes, and generate synthetic data with just a few keystrokes.
A new deep learning method, Point-Line Net, improves maize field phenotypic detection with high accuracy and efficiency. The model achieves an object detection accuracy of 81.5%, outperforming traditional methods in complex field environments.
The DEKR-SPrior model significantly reduces the mean absolute error in pod phenotyping compared to existing models, offering a valuable tool for enhancing crop yield predictions. It demonstrates superior accuracy in accurately detecting and counting soybean pods and seeds.
Researchers recreated the structure of supernova remnant SN 1181 using a new computer model, explaining its double shock formation. The study also found that high-speed stellar winds may have started blowing from its surface within the past 20-30 years.
Researchers develop a method that fuses AlphaFold's strengths with computer simulations based on physics laws to predict protein structures, enabling faster drug development. The approach filters down initial hypotheses to a more manageable set of structures, increasing the effectiveness of pharmaceuticals.
A newly developed deep learning algorithm has outperformed existing methods in predicting osteoporosis risk, potentially leading to earlier diagnoses and better outcomes. The model identified key factors such as weight, age, and grip strength as significant contributors to osteoporosis risk.
Researchers found that AI models that analyze medical images can predict patient demographics with high accuracy but struggle to diagnose patients from diverse backgrounds. The models use demographic shortcuts, leading to incorrect results for women, Black people, and other groups.
A new study from the University of Copenhagen reveals that AI has helped detect significantly more cases of breast cancer and reduce radiologist workloads. The AI system has been used to analyze tens of thousands of X-rays every year, resulting in a 12% increase in detected cases, including more small tumours of one centimetre or less.
A team of engineers has created a new mathematical model to accurately simulate the effects of blood flow on the adhesion and retention of nanoparticle drug carriers. The model, developed by University of Illinois professors Arif Masud and Hyunjoon Kong, was tested in vitro and demonstrated promising results.
A study found that large language models (LLMs) like ChatGPT underperform state-of-the-art detectors but can explain their analysis in plain language. LLMs' semantic knowledge makes them well-suited for detecting deepfakes, providing a common sense understanding of reality.
Researchers at Drexel University model the potential extent of flooding and combined sewer overflows in Camden, New Jersey, as climate change exacerbates the problem. The team's 'all-pipes' model simulates stormwater flows through every surface and pipe in the area, providing a unique solution to mitigate environmental and health risks.
Scientists developed a new approach to drive chemical reactions, generating compounds with unique pharmaceutical properties. The method uses photocatalysis and computational models to predict successful reactions, expanding the range of accessible substrates.
Researchers at Boston University developed an AI model that analyzes speech patterns to predict the likelihood of Alzheimer's disease in patients with mild cognitive impairment. The model achieved an accuracy rate of 78.5% and could potentially revolutionize dementia screening, making it more accessible and efficient.
A new method, PURPLE, estimates how often underreported health conditions occur in different demographic groups, providing a more accurate picture of intimate partner violence. The algorithm shows that women from lower-income brackets are more likely to experience violence, consistent with previous literature.
Researchers developed a symbolic model checking approach to verify quantum circuits, addressing the gap between model-checking quantum programs and quantum circuits. They used Maude programming language to formally specify and verify quantum circuits, confirming their correctness and paving the way for error-free quantum computing.
A new tool, BioemuS, enables real-time emulation and hybridization of biological systems using biomimetic Spiking Neural Networks, addressing limitations of current pharmacological treatments. The system, developed through international collaboration, prioritizes cost efficiency and accessibility for closed-loop applications.
A new model developed by researchers at the University of São Paulo combines physical parameters and machine learning to predict storm tides. The model uses physics-informed machine learning, which harmonizes physical models with measured data to produce more precise forecasts.
A recent Illinois-led study found that soil moisture variability remains consistent across growing and non-growing seasons in fields across the Midwest. The research team used sensor measurements and remote sensing data to reveal a stable pattern of dry and wet areas, which can be used to estimate high-resolution soil moisture products.
Researchers from MIT and the MIT-IBM Watson AI Lab devised a navigation method that converts visual representations into pieces of language, which are then fed into one large language model. The approach utilizes purely language-based representations, generating synthetic training data to overcome challenges in visual data availability.
Researchers from MIT develop a new technique called Natural Language Embedded Programs (NLEPs) to enable large language models to solve numerical, analytical, and language-based tasks. NLEPs achieved greater than 90 percent accuracy on symbolic reasoning tasks and showed 30 percent greater accuracy than task-specific prompting methods.
Researchers have developed a computer model that sheds light on extracting renewable energy from superhot, super deep rock. The model shows the formation of microscopic cracks creating a dense 'cloud of permeability' throughout the affected rock, which can lead to higher power delivery and efficiency.
Researchers discovered that brain regions, including anterior insula and striatum, are involved in the biased response to pleasure in people with bipolar disorder. In those with bipolar disorder, these brain regions show reduced communication, leading to a 'vicious cycle' of escalating mood and risk-taking behavior.
Dr. Jeetain Mittal's NIH grant will support multiscale computational models investigating phase separation in biology, particularly heterochromatin formation and its role in neurodegenerative diseases. The research aims to elucidate the molecular origins of phase separation using innovative models and methods.
Researchers have identified two distinct types of dysbioses in cystic fibrosis, which differ in their ecological organization and response to treatment. These findings suggest that antimicrobial drugs may be more effective in hierarchically organized microbiota, potentially leading to improved treatment outcomes for patients.
Astrophysicists calculate that two million years ago, the solar system encountered a cold, harsh interstellar cloud, which may have interfered with the sun's solar wind and affected Earth's climate. The heliosphere, a protective plasma shield, was compressed in such a way that it briefly placed Earth outside its influence.
A new method developed by researchers at Florida Atlantic University and the Smithsonian Environmental Research Center improves the accuracy of acoustic tracking in marine animals. The method uses a movement model to reconstruct animal tracks, even in regions with uneven receiver coverage.
A study by Dr. Edwin Dalmaijer found that pigeons' desire for social proximity leads to improved flight paths as younger birds learn from older ones. This generational improvement in route efficiency is similar to those seen in real-life data, suggesting a key role for social factors in navigation.
Six UTA faculty members have received prestigious CAREER grants from the National Science Foundation for their innovative research. The total award amount is $3.23 million.
A new AI model developed by researchers at the University of Jyvåskilö can predict and respond to human emotions, improving user experience. The model simulates cognitive evaluation processes to assess emotional responses to events, enabling computers to preemptively predict and mitigate negative emotions.
A new study from UCL reveals that Large Language Models exhibit irrational behavior, failing to reason logically and making simple mistakes. Despite their sophisticated capabilities, these AIs consistently fabricate information and respond inconsistently.
A computer model simulates dynamics of barrier island systems over two centuries, showing how human efforts to protect communities affect natural processes. The study finds that successful storm prevention strategies lead to less resilient barriers in the long term.
Researchers combined robotics data from different sources using generative AI models to train better multipurpose robots. The approach, known as Policy Composition (PoCo), enables a robot to perform multiple tool-use tasks and adapt to new tasks in unfamiliar environments.
Researchers from Mount Sinai presented three studies on obstructive sleep apnea (OSA) at SLEEP 2024, exploring its effects on thalamic activity, overnight memory performance, and symptom improvement with CPAP treatment. The findings suggest that OSA can impair cognitive function and that hypoxic burden may predict clinical improvement.