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.
A team of researchers has developed a DNA-based data storage platform with an expanded molecular alphabet, enabling the storage of vast amounts of digital information. The new system uses nanopores to distinguish between natural and chemically modified nucleotides, increasing storage density and sustainability.
A machine learning model has been developed to predict cognitive performance based on reaction time, eye movement, and difficulty level. The model achieved an accuracy of 82.8% in predicting participant success in visual attention tasks.
A new framework for portfolio management uses deep reinforcement learning to predict price trends and make strategic decisions, overcoming limitations of existing systems. The system consists of evolving agent modules and strategic agent modules, allowing for modular design and scalability.
Researchers at Mainz University will examine algorithm decisions on transparency, fairness, and data protection while optimizing resource use. The project aims to create workable trade-offs for applications and integrates young researcher promotion programs.
Researchers developed a machine learning model that provides good predictions for human speech recognition in noisy environments, benefiting hearing-impaired listeners. The model outperformed expectations and showed strong correlations with measured data.
The end-Permian mass extinction was characterized by a 10-degree climate warming, with 75% of organisms going extinct on land and 90% in oceans. Machine learning analysis reveals that declining oxygen levels, rising water temperatures, and ocean acidification were the key factors in organism survival or extinction.
Researchers at University of Illinois develop new method to accurately estimate soil organic carbon using airborne and satellite hyperspectral sensing. The study leverages machine learning algorithms with a comprehensive soil spectral library, enabling large-scale monitoring of surface soil organic carbon.
Researchers propose various approaches using AI, deep learning, and machine learning to improve the accuracy and predictive power of biomarkers for cancer and other diseases. The tools have shown promising applications in identifying early-stage cancers, inferring the site of specific cancers, and predicting response to immunotherapy.
Researchers found that ant colonies use an algorithm similar to the internet's data optimization, which senses and stabilizes behavior. This principle is also used in cells and neurons. Nature's algorithms may inspire new cybersecurity strategies or alternative approaches to gene regulation.
The SynGAP Research Fund has developed a pre-screening tool to identify potential SYNGAP1 patients through a free online survey. The partnership with Probably Genetic aims to screen undiagnosed patients and provide them with genetic testing resources, ultimately advancing treatment development for SYNGAP1.
A study using machine learning identifies climatic thresholds driving vegetation distribution, highlighting the importance of extreme climate conditions for savannas and deciduous forests. The findings provide valuable insights for improving process-based vegetation models and their coupling with Earth System Models.
Researchers developed a new deep learning algorithm that allows for real-time reconstruction of images combining optical and magnetic resonance imaging data. The algorithm, Z-Net, enables faster image generation and can be trained with simulated data, improving breast cancer detection.
A team of scientists developed a soft haptic sensor that can accurately estimate contact points and forces using computer vision and deep neural networks. The sensor is sensitive enough to detect even tiny forces and detailed object shapes.
Researchers developed a machine-learning technique that can pinpoint anomalies in large datasets, such as power grid failures and traffic bottlenecks. The model uses advanced probability distributions to identify low-density values, allowing for faster and more accurate anomaly detection.
Scientists have gained a new understanding of the atomic level interactions in complex catalysis, enabling more efficient and sustainable chemical production. Researchers used x-ray spectroscopy, machine learning analysis, and first principles calculations to model reactions and identify active site structures.
A new AI model combines multiple machine learning methods to accurately detect thyroid cancer and predict treatment outcomes from routine ultrasound images. The multimodal platform achieved high accuracy rates in detecting malignancies and predicting pathological stage and genomic mutations.
Researchers created a comprehensive genomic regulatory map of a 24-hour-old zebrafish embryo, identifying millions of regulatory segments that control gene transcription. The study used single-cell technologies and machine learning algorithms to analyze genome data from over 23,000 nuclei.
Pharmaceutical firms are working towards using machine learning to analyze vast stores of data, developing models that evolve and improve as the data are processed. However, experts agree that a fully functional end-to-end approach is still a ways off due to biology's complexity.
Researchers at the University of Illinois have developed a new type of water filtration membrane that mimics the natural process of morphogenesis. The membranes, made from soft polymers, exhibit complex 3D structures that allow them to efficiently separate pollutants from water.
A new study at McGill University found that an AI tutoring system improved surgical skills and learning outcomes among medical students compared to human instructors. The AI-powered Virtual Operative Assistant taught safe techniques and provided personalized feedback, leading to a 36% better performance rate.
Researchers used AI to predict flood damage in the US, finding a high probability of flood damage for more than 1.01 million square miles across the country. The study suggests that recent FEMA maps do not capture the full extent of flood risk, with 84.5% of reported damage not within high-risk flood areas.
The University of Essex team has devised a new approach to training neural networks called Target Space, which stabilizes the learning process by tweaking neuron firing strengths. This method enables deeper neural networks with fewer training examples and computing resources, accelerating AI breakthroughs.
Researchers studied how diverse neural network training datasets impact generalization. They found that data diversity is key to overcoming bias, but also degrade performance when neural networks are trained for multiple tasks simultaneously. The study highlights the importance of designing diverse and controlled datasets in machine le...
Researchers at the University of Copenhagen have developed an AI method to recognize and detect insect species based on their wingbeats, enabling easier monitoring of biodiversity. The method uses infrared sensors to measure wingbeats and group insects into different species without human input.
Researchers at Tokyo Institute of Technology have developed a new AI processor called Hiddenite, which achieves state-of-the-art accuracy in sparse neural networks with lower computational burdens. The chip drastically reduces external memory access for enhanced computational efficiency.
A new imaging method, combining optimal imaging and machine learning, can determine cell viability without chemical staining. The approach achieved a 95% accuracy rate and has the potential to be applied in hospitals and research labs.
Recent advances in AI for drug design have shown promising results, but further improvement is needed to translate early successes into effective drugs. Active learning and explainable AI hold the key to harnessing data value and designing correct molecules.
Jonathan Niles-Weed, an assistant professor at NYU's Courant Institute of Mathematical Sciences, has been awarded a prestigious Sloan Research Fellowship for his groundbreaking work in statistical theory and optimal transport. The fellowship recognizes his creativity, innovation, and research accomplishments in the field of data science.
A research team developed an algorithm that instantly assesses and adjusts brain stimulation placement using electroencephalography (EEG) feedback. The method can find the optimal stimulation parameters in just 1-2 minutes, potentially improving the efficacy of TMS treatment for brain disorders like chronic pain and depression.
A UCI team uncovered key brain mechanisms by which the hippocampus organizes memories into sequences, enabling decision-making. The finding may help understand memory failures in Alzheimer's disease and other forms of dementia.
A machine-learning algorithm named MAD3 can predict mechanical properties of metals without performing physical tests, cutting testing time by 1,000 times. The algorithm replaces traditional simulation software, enabling faster research and development with minimal resources.
Researchers at Mayo Clinic Cancer Center developed a machine learning algorithm that integrates genetic data from over 5,000 patients to predict patient benefit from chemotherapy and immunotherapy. A 32-gene molecular signature was identified, providing prognostic information and predicting patient response to immunotherapy.
Researchers found AI-synthesized faces to be nearly indistinguishable from real faces and rated as more trustworthy. The study's results have significant implications for the spread of manipulated images, including potential use in revenge porn and propaganda.
Researchers at Universidad Carlos III de Madrid developed a computer vision system to analyze cells in microscopy videos, allowing for automatic characterization of cell behavior. The system enables faster analysis of thousands of cells compared to traditional methods, which typically involve manual segmentation and tracking.
Anastasios Kyrillidis has won a National Science Foundation CAREER Award to explore the theory and design of non-convex optimization algorithms. His research aims to devise algorithmic foundations and theory that will accelerate problem-solving in machine learning, information processing, and optimization.
Researchers developed a new reagent-free detection technique for SARS-CoV-2 using Raman spectroscopy and machine learning. The method shows an accuracy of 80% in detecting COVID-19 infections from saliva samples, overcoming limitations of RT-PCR testing.
Researchers developed a hybrid machine-learning approach combining CNN and LSTM to recognize complex hand gestures in prosthetic hands. The technique achieved far superior performance than traditional machine learning efforts, with an accuracy of over 80%, but struggled with certain pinching gestures.
Researchers at Medical University of Vienna discovered that certain microbiome profiles following lung transplantation can provide prognostic information for future changes in lung function. Machine Learning analyzed multiomics data, including the microbiome, lipidome, and metabolome, to predict lung function deterioration.
Researchers developed a method called 6mASCOPE that measures DNA tagging system accuracy and distinguishes bacterial from human DNA. The study found high levels of methylation in plant, fly, mouse, and human cells, but mostly attributed to contamination.
Researchers developed a machine learning algorithm to automate propofol dosing for unconscious patients, matching human performance in sophisticated simulations. The 'dose penalty' model improved upon traditional software, but limitations remain, highlighting challenges in AI system accuracy and real-world application.
Researchers developed a neural network algorithm to predict art auction prices, relying on visual and non-visual characteristics of artworks. While human experts were more accurate, the algorithm showed promise in identifying market inefficiencies, such as auctioneer biases.
A new technique uses compression to reduce data transmission size, allowing for efficient federated learning on wireless devices. The approach has been shown to condense data packets by up to 99%, making it suitable for areas with limited bandwidth.
Rice University scientists employ machine-learning techniques to streamline the process of synthesizing graphene from waste through flash Joule heating. The lab used its custom optimization model to improve graphene crystallization from four starting materials over 173 trials.
Researchers developed a novel scheme that reduces energy consumption while improving traffic prediction accuracy by up to 40% compared to benchmark schemes. The scheme uses software defined network and edge computing to control the operation of base stations, ensuring both sustainability and performance.
Researchers developed MonoCon, a new AI technique that enables accurate identification of 3D objects in 2D images. By incorporating auxiliary context, the method improves object detection and estimation accuracy, paving the way for safer and more robust autonomous vehicles.
Researchers developed AI models trained on clinical data from a statewide health information exchange to predict healthcare resource utilization for individuals with COVID-19. The models demonstrated accurate public health predictions and provided valuable insights into patient-level need for healthcare resources.
A team of scientists developed an AI-based model to predict personal thermal comfort based on spatial parameters, achieving exceptional accuracy. The study highlights the importance of incorporating architectural features in models to reduce energy consumption.
Researchers at the University of Oklahoma have developed a molecular framework that solves the challenge of predicting peptide structures. The framework bridges experimental and computer sciences, enabling the use of machine learning and artificial intelligence to model peptide structures for materials engineering.
Researchers developed an AI system that can analyze retinal scans to identify patients at high risk of a heart attack over the next year. The system uses deep learning techniques and achieves an accuracy of 70-80%, revolutionizing the way patients are screened for signs of heart disease.
Scientists at Vienna University of Technology have developed a new type of neural network that can accurately simulate the quark-gluon plasma, a state of matter present in the early universe. The networks use gauge invariant convolutional neural networks to recognize patterns and predict properties of the plasma.
Physicists have detected X particles in quark-gluon plasma produced in the Large Hadron Collider, a phenomenon that could reveal the particles' unknown structure. The discovery uses machine-learning techniques to sift through massive datasets and identify decay patterns characteristic of X particles.
A team at the University of Washington has created an optical computing system that not only reduces noise but also utilizes it to improve creative output. The system uses a Generative Adversarial Network and demonstrates the viability of this technology at a large scale.
A new approach uses reinforcement learning algorithm to help robotic knee mimic intact human knee in walking, achieving 100% success rate on even ground. The technology also adapts to uneven terrain and changes in walking pace, promising a more comfortable experience for prosthetic users.
Researchers developed a machine learning approach enabling robots to separate, recognize, and grasp individual objects with high accuracy. The method achieved 97% success rate in real-world experiments, paving the way for industrial parts sorting and residential waste sorting applications.
A new federally funded study will use machine learning to create a first-of-its-kind algorithm predicting individual responses to food and dietary routines. The National Institutes of Health's All of Us Research Program will recruit 10,000 participants nationwide.
MIT researchers develop teaching phase that guides humans in understanding AI strengths and weaknesses, enabling more accurate decisions and faster conclusions. The technique helps humans build a mental model of the AI agent, reducing reliance on biased assumptions.
Researchers used electronic health record data from over 700 hospitals to train and evaluate three machine learning algorithms, finding that XGBoost provided the highest accuracy in predicting CDI among hospitalized patients. The study suggests that MLAs can help reduce the clinical and economic impact of healthcare-associated infections.
MIT researchers develop a method to test feature-attribution methods for machine-learning models. They find that even the most popular methods often miss important features in an image and some perform as poorly as a random baseline. This has major implications for high-stakes situations like medical diagnoses.
Research finds a strong association between retinal age and real age, with large gaps associated with increased mortality risk. The 'retinal age gap' could be used as a screening tool to predict mortality risk, particularly for non-cardiovascular causes.