Research using machine learning to solve scientific problems. Covers AI quality control, sea-surface temperature estimation, label-free cell analysis, and cyber defense.
CHOP researchers developed CancerVar, an artificial intelligence-empowered platform for interpreting somatic cancer mutations. The tool provides standardized procedures for assessing the clinical impacts of over 13 million somatic cancer mutations.
Researchers at the University of Missouri are applying AI to analyze protein dynamics, identifying potential target sites for new drug therapies. The approach can simulate protein changes related to conditions like cancer, enhancing the chances of successful therapies.
MIT researchers develop ExSum, a framework to formalize explanations of machine-learning models into quantifiable rules. This allows for testing assumptions about model behavior and reveals unexpected insights, such as negative words having sharper contributions to model decisions.
A new study uses artificial intelligence to predict bone fractures in cancer patients by creating a digital twin of the vertebra. The AI-assisted framework, ReconGAN, simulates how tumors affect the spine and predicts fracture risks, offering medical experts better treatment strategies and patient decisions.
Artificial Intelligence can now identify legendary batting techniques used by Sir Donald Bradman and modern players. Researchers developed a deep learning computer vision AI model to detect lateral backlift batters from straight ones.
Scientists have developed a machine learning algorithm that can accurately predict the lifetimes of different battery chemistries using as little as a single cycle of experimental data. The technique could reduce costs and accelerate the development of new battery materials, enabling researchers to quickly evaluate and test multiple ma...
A recent Dartmouth study using mobile technology compares emotional well-being of college students before and during the pandemic, finding significant differences in mental health and behavior. The research identifies two groups with differing experiences, with one group showing poorer mental health, anxiety, and stress.
Researchers at Caltech developed Neural-Fly, a deep-learning method that enables drones to adapt to wind conditions in real-time. The method achieved significant improvements in drone performance compared to existing adaptive control algorithms.
Researchers at DTU Compute and DIKU have developed a machine learning model that can map the potential of proteins, enabling the biotech industry to accelerate the development of new proteins. The model generates a picture of how proteins are linked, allowing for the identification of closely related proteins with desirable properties.
Researchers from Dartmouth College used artificial intelligence to draft wine and beer reviews, finding agreement between human and machine-generated reviews. The team also developed a system to write review syntheses, aggregating elements from existing reviews to provide limited but relevant information about products.
Researchers from Bar-Ilan University discovered that brain learning occurs mainly in dendritic trees, where the trunk and branches modify their strength. This finding paves the way for a new type of AI hardware and algorithms with comparable success rates to existing parallel GPUs.
Agricultural nitrous oxide emissions are estimated to be 300 times more powerful than carbon dioxide in trapping heat. The new knowledge-guided machine learning model, KGML-ag, is 1,000 times faster and more accurate than current systems, providing a promising solution for reducing greenhouse gas emissions from agriculture.
Researchers developed an enzyme that can break down plastic waste in hours, making it a promising solution for the world's plastic problem. The enzyme, called FAST-PETase, has the potential to revolutionize recycling and reduce global landfill waste by billions of tons.
Researchers at MIT developed an AI method that constrains machine-learning models to suggest molecules with producible chemical structures. The approach guarantees quality and speed, outperforming existing methods in proposing high-quality molecular structures.
Researchers from the University of Cambridge have developed a hybrid, data-driven approach to the Travelling Salesperson Problem that significantly outperforms current methods. The new approach converges to optimal solutions at a faster rate and produces high-quality solutions under restricted computation time.
A team of researchers developed a prognostic model using automated brain scans and machine learning to predict survival and recovery in severe TBI patients. The model accurately predicted risk of death and unfavorable outcomes at six months after the injury, potentially improving clinical decision-making and patient care.
Scientists at the University of Oxford have developed an 'optomemristor' device that facilitates three-factor learning and emulation of biological computations, making it possible to perform complex machine learning tasks. The device uses both light and electrical signals to interact and consume very little energy.
Researchers developed a machine-learning method that allows robots to pick up and place never-before-seen objects in random poses, requiring only 10 human demonstrations. The system uses a neural network specifically designed to reconstruct 3D shapes, enabling the robot to generalize to new object orientations.
Researchers developed an AI system using swarm learning to predict cancer from medical images of tissue samples without accessing patient data. The technique improved the detection of genetic changes in colon tumors with high accuracy.
Researchers developed a new decision-support tool to optimize ship channel dredging and disposal activities, considering factors like navigability condition deterioration and economic values. The algorithm weighs input from local professionals and projects costs with interest and inflation included.
Deep learning is being adopted in optical metrology to solve ill-posed inverse problems, such as model mismatch and error accumulation. This approach leverages large datasets and active control strategies to produce accurate reconstruction results.
A new control allocation method using a neural network improves the performance of quadrotor controllers by considering aerodynamic effects. This approach reduces errors in command generation and delivers better thrust and torque signals.
A University of Arkansas professor is developing learning algorithms for building fair decision models in both offline and online learning settings. He will utilize Pearl's Structural Causal Model to analyze causal effects from observational data and propose universal formulations for measuring long-term fairness.
The new computational tool, AF2Complex, predicts the structure of protein complexes and their interactions, offering insights into biomolecular mechanisms. The model is based on AlphaFold 2 and performs well in predicting protein structures and complex formations.
A new KAUST study uses machine learning to predict disease spread with high accuracy, dynamically incorporating latest data without human bias. This approach offers a promising alternative to conventional models, providing a more accurate story of the underlying epidemic data.
A novel method developed by the University of Tsukuba uses drones and machine learning to estimate the amount of plastic litter in rivers. The approach combines high-resolution optical and thermal images, resulting in more accurate estimates than other methods.
A study found that trainee teachers who received AI-generated feedback improved their diagnostic reasoning, identifying potential learning difficulties in pupils more accurately. The AI system analyzed the trainees' work and provided clear, adaptive feedback.
Scientists develop models that complement simulations using reinforcement learning and numerical methods to predict climate change, turbulent flows, and morphogenesis. This approach enables faster and more energy-efficient predictions, solving complex problems in engineering and climate applications.
A machine learning model has been trained to accurately identify individuals with post-traumatic stress disorder (PTSD) by analyzing text data. The model achieved an 80% accuracy rate in distinguishing between those with and without PTSD. This breakthrough could lead to the development of a cost-effective screening tool for health prof...
Researchers used AI to analyze videos of over 2,500 ASL signs and found that challenging signs are made closer to the signer's face, making them easier for perceivers to recognize. This suggests that ASL has evolved to be more recognizable, improving communication.
A proof-of-concept study demonstrates that machine learning technology can enable humans to see in the dark with full-color night vision. The research uses deep learning to enhance color perception in low-light conditions.
Researchers have developed a new method called Shared Interest that enables users to aggregate, sort, and rank individual explanations of a machine-learning model's reasoning. This technique uses quantifiable metrics to compare how well the model's reasoning matches human thinking, helping to uncover concerning trends in decision-making.
A novel 'rational' neural network reveals underlying mathematical equations through Green's functions, enabling humans to understand machine-generated findings. This breakthrough in partial differential equation learning holds promise for advancing scientific exploration of weather systems, climate change, and more.
A new Stanford University-led study uses machine learning and human insights to map regions and ports most at risk for illicit practices, like forced labor or illegal catch. The results highlight two main risk factors: the vessel's flag state and type of fishing gear onboard.
A new e-nose prototype, NOS.E, can distinguish between six whiskies by brand names, regions, and styles in under four minutes, with 100% accuracy for region detection and 96.15% for brand name identification. The technology has applications beyond whisky, including counterfeiting detection in perfume and wine.
Researchers found that AI-enhanced diagnosis helps doctors accurately detect fetal congenital heart disease, with fellows making the most accurate diagnoses. The new system uses graphical charts to represent the AI's analysis of ultrasound videos, improving accuracy and trust among medical professionals.
Researchers at MIT developed a framework for robotic manipulation systems that can perform complex tasks using a two-stage learning process. This allows robots to learn abstract ideas about manipulating deformable objects, such as pizza dough, and execute skills to complete tasks.
Researchers integrated biological signals with gold-standard machine learning methods to create emotionally intelligent speech dialog systems. The study found that combining language information with biological signal information improved the AI's performance, making it comparable to human-like emotional recognition.
Researchers at MIT created a process called DualFair that can remove bias from data used to train machine-learning models. The method tackles both label bias and selection bias, significantly reducing discrimination in loan predictions while maintaining high accuracy.
The VIGICOVID system uses deep neural architectures to answer natural language questions about COVID-19 and SARS-CoV-2. Researchers have developed a prototype that provides good results and can be easily scaled up for marketability.
Recent advances in machine learning (ML) and artificial intelligence (AI) have revolutionized the field of metaphotonics. The integration of ML with photonics enables the creation of intelligent systems that can adapt to changing environmental conditions. Self-adapting systems, such as cloaks that adjust themselves to changes in freque...
Researchers developed an AI-driven image analysis pipeline that identified novel cellular hallmarks of Parkinson's disease from images of over a million skin cells. The platform can distinguish between patient cells and healthy controls, revealing new signatures for potential therapeutic targets.
Researchers developed an AI method to analyze electronic health record data and predict optimal drug regimens for type 2 diabetes patients with similar characteristics. The algorithm successfully supported medication selection for over 83% of patients, leading to better management of the disease and improved patient engagement.
Researchers developed a lightweight exoskeleton that uses machine learning to predict user intentions and provide assistance. The system successfully helped participants stand up, demonstrating potential for supporting individuals with mobility impairments.
Researchers warn of machine learning bias when data published for one task is used to train algorithms for a different one. This can lead to compromised integrity and 'overly optimistic' results in medical imaging applications.
Engineer Thomas Senftle at Rice University has won a prestigious NSF CAREER Award to improve catalysts through machine learning. He will develop open-source models to speed up the development of catalysts with optimized particle/support combinations, aiming to reduce unwanted molecules in water.
Researchers developed an AI-powered model to assess rare-earth compound stability, leveraging machine learning and high-throughput density-functional theory. This framework has far-reaching applications in materials science, including designing new compounds for clean energy technologies and optimizing magnetic properties.
A new study from St John's College, University of Cambridge suggests that robots can help produce solar fuels, accelerating the world's transition to green renewables. The 'cyber-leaf' concept uses AI and robots to create sustainable syngas, reducing reliance on fossil fuels.
A new AI method accurately estimates timetable robustness within milliseconds, enabling efficient optimization. This improves balance between passenger needs and economic conditions.
Researchers at NC State University have developed a 'self-driving lab' that uses artificial intelligence and fluidic systems to advance our understanding of metal halide perovskite nanocrystals. The technology can autonomously dope MHP nanocrystals, adding manganese atoms on demand, allowing for faster control over properties.
Scientists developed an AI approach to model and map the Earth's natural features in greater detail and accuracy. The new system can recognise intricate features and aspects of the terrain far beyond traditional methods, generating enhanced-quality environmental maps.
A team of researchers from McGill University and other institutions analyzed 6,850 testimonials from psychedelic users, creating a 3D map of brain receptors linked to subjective experiences. The study found associations between serotonin receptors and ego-dissolution, suggesting potential new treatments for psychiatric conditions.
Researchers at Duke University found a collection of coordinated brain regions that predict and direct social behavior in mice. By analyzing the electrical activity of these regions, they identified how social or solitary an individual mouse is and were able to prompt them to be more gregarious. This study may lead to better diagnostic...
A study published in Frontiers in Microbiology has found that machine learning analysis of microscopy images can be used to identify bacteria resistant to antibiotics. Researchers discovered that shape changes in bacterial cells can predict drug resistance, suggesting a new approach for detecting and predicting drug resistance.
A Bocconi University study used machine learning to analyze data on 2038 couples in Germany, finding that life satisfaction and housework are key predictors of union dissolution. The analysis also revealed complex interactions between variables, including the impact of personal traits like openness and extraversion.
Researchers using machine learning methods risk underestimating uncertainties in their final results due to decorrelation with imperfections in simulations. This could weaken or bias classifier algorithms' ability to identify fundamental particles.
Researchers at Texas A&M University developed a novel error estimator using transfer learning principles to evaluate machine-learning model performance. The technique enables fast screening of source data sets, improving the accuracy of diagnoses in complex medical issues like schizophrenia.
Researchers at West Virginia University are using machine learning and geographic information systems to identify areas with low COVID-vaccine uptake. They aim to pinpoint counties with increased risk of outbreaks, predict where testing is most crucial, and develop targeted interventions to increase testing rates. By acknowledging comm...
Researchers in Brazil developed a computer program that locates swimming pools and rooftop water tanks in aerial photographs, using artificial intelligence to identify socio-economically deprived urban areas at risk for diseases transmitted by Aedes aegypti. The innovation can be used as a public policy tool for dynamic socio-economic ...
A new mathematical model developed by UCI researchers combines human and algorithmic predictions and confidence scores to improve AI accuracy. The hybrid model outperforms individual human or machine predictions, demonstrating the potential of human-AI collaboration in building smarter AI systems.