Articles tagged with Algorithms
A new AI tool developed by researchers at Stanford University improves clinicians' ability to correctly identify brain aneurysms by highlighting areas of interest on scans. The HeadXNet algorithm reduces the 'miss' rate and increases consensus among clinicians, with promising results but further investigation needed.
A new microgrid system designed by American and Chinese researchers promises improved stability, safety, and resilience in delivering energy. The system aligns several energy sources in parallel and uses a decentralized control algorithm to overcome the burdens of system overload and shutdown.
Researchers at RIT are developing a computational framework to reconstruct and visualize myocardial active stress, enabling direct appraisal of cardiac function. This will help detect and localize regions with diminished contractile activity, improving diagnosis and treatment of cardiac disease.
Researchers develop a new algorithm using over 230,000 patient records to predict mortality risk and survival chances. The algorithm uses disease history from the past 23 years to make more accurate predictions than existing methods.
Researchers at Pitt and CMU aim to develop AI-powered robotic suits that can treat and stabilize soldiers in the field, extending the 'golden hour' for medical evacuations. The project uses real-world data and medical algorithms to guide critical care interventions.
USC researchers Bhaskar Krishnamachari and Aditya Asgaonkar propose a blockchain-based solution to reduce costs and increase efficiency in online transactions. Their system allows buyers and sellers to interact directly without third-party mediators, using smart contracts to verify honesty and disincentivize malicious behavior.
A new algorithm called DeeProtein uses sensitivity analysis to unravel the secret of its predictions, providing valuable insights into protein functions. This technique enables researchers to identify critical regions in proteins that tolerate changes well or poorly, paving the way for targeted modifications.
A new brain-controlled hearing aid technology developed by Columbia engineers can identify and amplify the correct speaker in a crowded environment. The device uses artificial intelligence to monitor wearers' brain waves and boost the voice they want to focus on, solving the 'cocktail party problem' that modern hearing aids struggle with.
Researchers developed an algorithm called Scanorama that merges over 20 diverse human cell datasets into a single, diverse source of data. The algorithm uses a modified computer-vision technique to find matching cells across datasets and preserves unique cell types.
A machine learning algorithm can detect signs of anxiety and depression in young children's speech, potentially providing a fast and easy way to diagnose conditions that are difficult to spot. The algorithm is highly successful at diagnosing children with an internalizing disorder with 80% accuracy.
A team of researchers has developed a mathematical model to calculate the cost - time and energy - to complete a task based on the number of drones and recharging stations available. The model considers the energy required for each drone to complete its portion of the task and fly to a charging station as needed.
Researchers have developed an energy renormalization algorithm to predict glass' mechanical behavior at varying temperatures. This approach enables the design of dynamic materials with optimal properties, scaling molecular simulations up by roughly a thousand times.
Researchers at Rutgers University have developed a low-cost, high-speed algorithm to test chemical toxicity without animals. The algorithm uses massive amounts of data from PubChem and achieves a 62-100% success rate in predicting oral toxicity for several groups of chemicals.
Researchers propose a unified approach to optimize radar polarimetry data for precise weather forecasting. This combines observation-based retrievals with model-based analysis to improve quantitative precipitation estimation, warnings, and forecasts.
A new AI approach analyzes time-lapse images of embryos to predict pregnancy potential, improving IVF success rates and minimizing multiple pregnancies. Researchers used a large dataset of human embryos to train an algorithm that can classify embryo quality with high accuracy.
A new mathematical method developed by USC researchers can identify anomalies in self-driving cars' perception algorithms before they hit the road, improving safety. The method uses 'sanity conditions' to test machine learning tools and can be used to pinpoint specific problems and retrain the algorithms for faster error detection.
A team of healthcare data scientists and doctors developed an AI system using machine learning algorithms to predict the risk of early death due to chronic disease. The system performed better than current standard approaches and showed promise in improving preventative healthcare.
Researchers used an algorithm to model the decision-making of C. elegans in response to a sensory stimulus, achieving predictions that matched experimental results. The Sir Isaac platform demonstrated improved accuracy compared to prior models, offering insights into the potential of artificial intelligence in scientific discovery.
The Blue Brain Project has developed an algorithm to objectively classify the shapes of neurons in the brain, enabling the creation of a standardized taxonomy of all brain cells. This breakthrough resolves a century-old neuroscience problem and provides a reliable comparative method for researchers.
Researchers at MIT have developed a neural architecture search algorithm that can directly learn specialized convolutional neural networks for target hardware platforms in only 200 GPU hours. The algorithm uses 'path-level' binarization and pruning to reduce memory consumption and improve efficiency.
Researchers at University of California, Davis and Maynooth University created programmable DNA molecules that can self-assemble into patterns by running their own program. They designed and ran 21 algorithms, demonstrating the potential of the system for sophisticated molecular engineering.
Researchers design DNA molecules that can carry out reprogrammable computations, executing six-bit algorithms for simple tasks. The system demonstrates versatility and potential for teaching about molecular processes encoding information.
Researchers from Mitsubishi Electric Research Laboratories developed an improved algorithm to track motor performance and speed estimation without sensors. The proposed algorithm uses state variables to estimate rotor speed, addressing limitations in existing approaches.
A new study published in Ophthalmology shows that AI can assist physicians in accurately diagnosing diabetic retinopathy, a potentially blinding eye disease. The researchers developed two types of assistance to help physicians read the algorithm's predictions, which improved diagnostic accuracy and confidence.
Researchers at Garvan Institute of Medical Research have developed a computational method to reduce the amount of memory necessary for genome alignment, allowing for real-time analysis on smartphones. This breakthrough enables remote disease identification and point-of-care microbial infections.
Researchers have created a novel way to define individual protein associations using the topological scoring (TopS) algorithm. This approach helps identify proteins that come together in biological processes, allowing for better understanding of how proteins perform functions and interact with each other.
Omri Weinstein, assistant professor at Columbia University, has won a National Science Foundation CAREER Award to explore data structures and information retrieval. His research aims to develop new mathematical tools for proving lower bounds on data structure efficiency and scalable storage applications.
Researchers from the Center for Genomic Regulation developed a method to predict and classify these tiny proteins using bioinformatics tools, discovering they account for 16% of bacterial genomes. The small proteins play a crucial role in antimicrobial responses, microbiota balance, and may be overlooked in complex organisms.
NIST researchers demonstrate deep learning algorithms outperform traditional methods for detecting offshore radars, improving spectrum sharing. The new approach provides occupancy statistics for the 3.5 GHz band, enabling commercial users to determine when to yield to naval operations.
Rice University researchers have developed a method to capture 4D data using 2D microscopes, enabling scientists to visualize molecules' locations and movements in living cells. The technique uses custom phase masks to manipulate light and separate spatial and temporal information.
A University of Texas at Arlington researcher is leading a high-powered team in the Advanced Research Projects Agency-Energy (ARPA-E) Grid Optimization Competition. The team has secured $250,000 in funding and aims to develop software solutions for efficient energy delivery through the nation's power grid.
A machine learning algorithm has been developed to speed up the process of discovering new medicines, identifying four new molecules that activate a protein relevant to symptoms of Alzheimer's disease and schizophrenia. The algorithm is twice as efficient as industry standards and can analyze vast amounts of chemical data.
A new study finds that women's brains are metabolically three years younger than men's of the same age, which may contribute to their greater mental sharpness in later years. The researchers used PET scans and machine-learning algorithms to measure brain metabolism and calculate each person's brain age.
Osaka University researchers developed an algorithm for numerical calculation of EM noise in electric circuits, reducing interference caused by transmission lines. The new method allows for more practical calculations and demonstrates the reduction of EM noise using symmetric 3-line configurations.
MIT engineers have developed an algorithm that enables autonomous underwater vehicles to weigh the risks and potential rewards of exploring unknown regions. The algorithm assesses risk levels and reward probabilities in real-time, allowing AUVs to take calculated risks when justified by potential scientific rewards.
Researchers developed an AI-powered method to correct errors in single-cell RNA sequencing, enabling precise data for every cell. The algorithm, called kBET, quantifies differences between experiments and facilitates comparison of correction results.
A novel model developed by MIT and Microsoft researchers identifies instances where autonomous systems learn from examples that don't match the real world, leading to dangerous errors. The model uses human feedback to pinpoint situations where more information is needed to act correctly.
Researchers developed a machine learning algorithm to optimize lab test ordering in ICU patients, reducing testing by up to 44% and improving treatment timing. The approach aims to maximize patient rewards while minimizing costs and risks.
Dr. Eric Baumer aims to develop participatory methods for human-centered design of algorithmic systems, incorporating diverse experts and users in the design process. He will work with nonprofits AEquitas and ProPublica to create interactive tools that better align with users' existing practices.
A recent study from the University of Waterloo found that measuring AI's ability to learn is challenging due to the complexity of tasks. The researchers discovered that no mathematical method can determine whether an AI-based tool can handle a task or not, even with precise task descriptions.
The Mathematical Association of America (MAA) recognizes Tom Leinster for his outstanding expository article 'Rethinking Set Theory' with the MAA Chauvenet Prize. Cathy O'Neill wins the MAA Euler Book Prize for her book 'Weapons of Math Destruction', tackling data science's social and political implications.
Machine learning helps detect epigenetic features in genomes and identifies similarities between phenotypes and gene modifications.
Researchers at University of Notre Dame have developed a new mathematical approach to solve NP-hard problems using analog computing. The 'solver' has the potential to find better and possibly faster solutions than digital computers for complex optimization problems.
A new algorithm that combines experimental data with machine learning reduces the time needed to find optimal peptide sequences, allowing for faster discovery and synthesis. This method has the potential to revolutionize how peptides are designed and could lead to breakthroughs in materials science, chemistry, and medicine.
Researchers at RIT are developing an advanced visual tracking system using deep learning and artificial intelligence to refine object location and movement. The system has potential applications in autonomous navigation, drones, traffic monitoring, safety, security, disaster response, and human-computer interaction.
Researchers at the Salk Institute discovered a framework that mimics how fruit flies detect novel odors, using a Bloom filter-like data structure. This new approach improved accuracy for detecting duplicates or anomalies in large datasets.
A novel scale-free mechanism guides the search of an artificial bee colony algorithm, improving its exploitation ability and maintaining population diversity. This enables a better balance between exploration and exploitation, leading to enhanced search ability in real-world optimization problems.
Researchers used network analysis to rank films and their creators based on references in subsequent movies. The study found that influential films were mostly produced before 1980 and featured top US directors like Hitchcock, Spielberg, and Kubrick.
A new algorithm computes and enables realistic movement and distortion of 3D surfaces, including gravity, contact, and friction. The method allows users to freely design garments around a 3D character without worrying about physics.
A RIT researcher is developing new signal processing solutions to improve data analysis efficiency and reliability. The project aims to reduce the impact of faulty measurements in complex sensing systems by creating algorithms that can detect and mitigate corrupted data.
Researchers have developed a new method to replicate complex natural tessellations on surface meshes using layered fields and sparse linear algebra kernels. The algorithm is platform-independent, concise, and achieves considerable performance gains over existing serial Voronoi diagram codes.
A new machine learning algorithm combined with a handheld smartphone device can accurately estimate the gestational age of premature newborns. The technology has the potential to aid healthcare workers in remote and low-income countries, where preterm births are a leading cause of child mortality.
Researchers identify potential causes of bias in machine learning systems and demonstrate how changing data collection methods can reduce bias without compromising accuracy. They suggest identifying clusters of patients with high disparities in accuracy to inform data collection decisions.
A team of researchers is using a $820,000 DARPA grant to design a new kind of chip that can automatically match software with hardware, significantly reducing the time it takes. The chip, called DASH-SoC, aims to achieve this in just five nanoseconds.
The Orcasound project has developed a web application that enables citizen scientists to listen to livestreaming audio from hydrophones, complementing computer algorithms in analyzing data. The app aims to bring synergy between human listeners and sophisticated algorithms, saving audio data to online cloud storage for later analysis.
A University of Pittsburgh researcher is using video games to test and improve artificial intelligence algorithms. The goal is to create AIs that can learn from their mistakes in complex, uncertain environments. By analyzing gameplay data, the algorithm can refine its strategies for optimal decision-making.
Researchers developed a new approach to interpret machine learning algorithms, revealing how they can produce nonsensical answers even when given meaningful inputs. By reducing inputs to the bare minimum required for correct answers, the method provides insights into algorithmic limitations and potential solutions.
Researchers at Dartmouth College used the Bible to develop an algorithm that can convert written works into different styles for different audiences. The study, published in Royal Society Open Science, trained on over 1.5 million unique pairings of source and target verses from various versions of the sacred texts.
A study published in EPJ Data Science found that algorithms predicting crimes more accurately than existing models, with significant improvements in predictions for assaults and drug offences. The system used location and activity data from Foursquare users to make predictions.
A computational tool using machine learning and optimization algorithms matches refugees to host communities with the best resources for success. Annie MOORE predicts refugee employment likelihood and guides resettlement workers in finding suitable placements, improving integration outcomes.