Articles tagged with Algorithms
Scientists used fNIRS to analyze brain activity and develop an algorithm to predict autistic traits severity with over 90% accuracy. The new technology is more accessible than fMRI for diagnosing autism spectrum disorders.
A new system called SpAtten enables more streamlined NLP with less computing power, achieving speeds of over 100 times faster than competing general-purpose processors and reducing energy consumption by over 1,000 times.
A recent study found that Google Scholar systematically relegates non-English language documents to invisible positions, even among quality articles with hundreds of citations. This bias can have detrimental effects on researchers from non-English speaking countries, leading them to believe there is no literature in their native language.
A machine-learning algorithm developed at UT Southwestern estimates that the number of COVID-19 cases in the US is nearly three times that of confirmed cases, with over 71 million people estimated to have contracted the virus. The algorithm provides daily updated estimates of total infections and the number of people currently infected...
Researchers at Geisinger Health System developed an AI algorithm using echocardiogram videos to predict mortality within a year. The model outperformed other clinically used predictors and improved cardiologists' prediction accuracy by 13 percent.
Researchers from HSE University developed a new method to process MEG data, enabling cortical activation areas with higher precision. This improvement allows for better diagnosis of neurological disorders and preparation for brain surgery.
Researchers from RUDN University found a way to reduce the size of a trained neural network by six times without retraining, achieving significant storage volume reduction and minimal accuracy loss. The new method leverages correlations between initial and simplified weights, eliminating the need for post-training.
The Ramanujan Machine generates mathematical conjectures without proof, imitating Ramanujan's intuition using AI and computer automation. It has already produced known formulas for pi, Euler's number, and other constants, as well as several unknown conjectures.
Researchers from Universitat Rovira i Virgili and Institute of Materials Science of Barcelona have combined experimental data with algorithms to enable an unprecedented predicting capability of the performance of organic solar cells. The study uses a new experimental method to generate large datasets, which are then used to train machi...
Machine-learning algorithms are trained on large datasets to predict the performance of organic solar cells, identifying key parameters such as electronic gap and charge transport balance. This study demonstrates a new approach to predicting material efficiency, paving the way for further analytical models and more complex system under...
Researchers at Texas A&M University developed a deep-learning algorithm that can denoise images to reveal otherwise invisible details. The algorithm, called global voxel transformer networks (GVTNets), uses adaptive receptive fields to capture information in the overall image structure.
Researchers have detected bias in face recognition algorithms, with higher false positive rates for females with dark skin tone and males with light skin tone. Top winning solutions exceeded 99.9% accuracy, but the analysis of top 10 teams showed that overall accuracy is not enough when building fair face recognition methods.
A new algorithm called Meta-Apo reduces the need for expensive whole-genome sequencing to understand microbial function, improving consistency with 16S rRNA gene amplicon results. This approach enables accurate diagnoses like gingivitis improvement from 65% to 95% using low-cost 16S-amplicon sequencing.
Researchers develop an innovative algorithm inspired by weakly electric fish to detect and locate objects via electrosensing. The multi-scale approach combines information gathered at different distances from the object, providing a more accurate understanding of its features.
Researchers at Università di Trento developed RoomTetris algorithm to efficiently allocate rooms, increasing hotel occupancy rates and profits. The software has shown promising results in tests, with potential to double profitability for some hotels.
Researchers developed fish-inspired robots that synchronize movements in 3D space, exhibiting complex collective behaviors such as aggregation and circle formation. The system uses blue LED lights for vision-based coordination and demonstrates autonomy in underwater environments.
Researchers demonstrate that no single algorithm can determine whether a superintelligent AI would produce harm, and deciding on its intelligence is incomputable. The study's containment problem highlights the challenges of controlling such an AI.
A Sandia Labs research team used machine learning to complete materials science calculations 42,000 times faster than normal, accelerating the creation of new technologies for optics, aerospace, and energy storage.
Researchers at Columbia University are developing a novel platform to enable factory workers to work remotely using a robot teleoperation system. The system will allow workers to specify high-level task goals without expert knowledge of the robot hardware and configuration, enabling greater job access for workers regardless of geograph...
Researchers at the University of Copenhagen find that computer algorithms cannot optimize packing of objects in two dimensions without overlap beyond four or five items. The study's findings have significant implications for industries such as clothing manufacturing and metal processing, where efficient material cutting is crucial.
Researchers use single-cell RNA sequencing to analyze kidney cells' gene expression, enabling the reconstruction of their spatial arrangement and functional information. The algorithm succeeds in completing a 3D puzzle, providing new insights into kidney disease.
Algorithms can adopt collusive pricing rules without human intervention, posing a threat to consumers. A multidisciplinary policy design is proposed to investigate AI pricing algorithms for collusion in controlled environments.
A research team from RUDN University created an algorithm to help large groups of people make optimal decisions quickly. They applied it to a real-life example, where the outbreak of COVID-19 required the administration and sellers to agree on compensation amounts within just three steps.
Researchers developed an AI algorithm called CAMEO that discovered a new compound by operating in a closed loop, maximizing productivity and efficiency. The AI is designed to contain knowledge of key principles, including past simulations and lab experiments, to identify the best material for specific applications.
Scientists are studying the infrasound signatures of tornadoes to develop more accurate prediction and warning systems. Researchers have found that these vibrations can travel long distances quickly and through different media, potentially allowing for early detection of tornadoes from far away.
A team of scientists from Tokyo Institute of Technology devised a strategy to automate the novel material development process using robotics and artificial intelligence. The CASH setup enables fully autonomous materials research, making it possible to test and optimize new compounds at an unprecedented scale.
Researchers developed a computer algorithm that captures traffic network topology to identify congestion hotspots. The algorithm suggests ways around congested areas in cities.
Researchers employed a computer algorithm to analyze nearly 15,000 landscape paintings, finding that compositional structures evolved systematically over time. The study suggests a potential bias in art curators' and historians' selections, contradicting the prevailing view of diverse artistic expression.
A new algorithm helps track fast charged particles in plasma, which could influence fusion reactions. The algorithm conserves energy during pitch-angle scattering, a critical process in fusion plasma.
A new open-source tool, CRYLOGGER, analyzes thousands of Android apps to detect cryptographic misuses without requiring access to the app's code. The study found that nearly all popular Android apps contain code or use libraries that do not strictly adhere to security standards.
A new machine learning algorithm has been developed to isolate dynamic neural patterns in brain signals that relate to specific behaviors, such as finger movements. The algorithm resolves the challenge of isolating these patterns, which can be masked by other activities, and enhances the decoding of behaviors from brain signals.
Researchers from the University of Michigan found that the vast majority of field massive stars in the Small Magellanic Cloud are 'runaways,' or stars ejected from clusters. They discovered that these stars could have formed in isolation or were dynamically ejected due to unstable orbital configurations.
Purdue University is leading a $3.7 million research project to create more secure machine learning algorithms for autonomous systems. The goal is to develop a robust, distributed and usable software suite to prevent AI hacking and ensure the accuracy of autonomous machines on the battlefield.
A new algorithm could help reduce stress on health systems in the UK and beyond by redistributing ICU patients across hospitals. The load balancing method, developed by Queen Mary University of London, has been shown to enable access for up to 1000 additional cases in the UK.
A team of researchers from Rensselaer Polytechnic Institute developed a deep neural network that can perform nearly as well as more complex dual-energy CT imaging technology. The algorithm produced high-quality approximations with a relative error of less than 2% using single-spectrum CT data.
Researchers found that AI techniques are being used to process information about brain structure and connectivity, assess surgical candidacy, and predict disease trajectory. However, they emphasize the need for explainable algorithms to ensure trust in algorithmic prescriptions or diagnoses.
Researchers created an abstract language that describes protein molecules' shapes and structures, enabling predictions of their dynamics. This method uses machine learning algorithms to analyze molecular movements and provides insights into disease causes and targeted drug therapies.
Researchers developed an algorithm to predict sex-specific adverse drug effects, identifying 20,817 adverse drug effects posing sex-specific risks. The algorithm addresses confounding biases and provides an opportunity to minimize adverse events by tailoring drug prescription and dosage to sex.
Skoltech researchers propose a fast and accurate numerical method to address the low spatial resolution of EEG studies. The new approach directly backpropagates measured signals from the skin down to the cortex, increasing accuracy for source localization while speeding up processing.
Cyber-physical systems face challenges due to large volumes of data flowing through communication networks, causing routing and queuing delays that degrade system quality. The new algorithms developed by researchers strike a balance between communication sparsity, delay, and performance, ensuring safe and stable operation.
The University of Texas at Austin has been selected to lead the NSF AI Institute for Foundations of Machine Learning, aiming to develop new classes of algorithms for more sophisticated AI technologies. The institute will focus on addressing challenges such as noise and diffuseness in modern datasets.
A new machine learning algorithm confirms 50 new exoplanets, ranging from Neptunes to Earth-sized worlds, by distinguishing between real and false positives. The technique outperforms previous methods, can be automated, and improves with further training.
The method can be used in a portable, tabletop device to rapidly identify known and emerging opioid fentanyl substances, aiding in the safety of law enforcement and military personnel. The AI algorithm had a 92.5 percent accuracy rate for correctly identifying molecules related to fentanyl.
Researchers have developed an efficient method to estimate camera movement, reducing the number of hypotheses generated from up to five to one. The new approach allows for real-time execution of pose estimation, with a complete algorithm taking only 29 milliseconds per frame.
Scientists at NIST have found a way to significantly enhance the accuracy of key information on how heat affects the stability of folded DNA structures. The novel mathematical algorithm automatically accounts for unknown effects, allowing scientists to design durable and complex structures made from DNA.
A new tool, FairCo, was developed to improve the fairness of online rankings by giving equally relevant choices roughly equal exposure and avoiding preferential treatment for high-ranked items. This can correct unfairness in existing algorithms and curtail personal choice.
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.
A new calibration-free version of a physical-based method can better detect long-term trends in continental-scale land evaporation rates than recent remote sensing-based approaches. This improvement is crucial for upgrading general circulation models' existing evaporation estimation algorithms to produce better climatic predictions.
Researchers Xifeng Yan and Yu-Xiang Wang developed a deep learning tool to forecast COVID-19 trends by community. The model, called Transformer, uses attention mechanisms and U.S. Census data to provide hyper-local forecasts with high accuracy.
The EBRAINS human brain atlas is a comprehensive digital map of the cellular architecture, showcasing 250 structurally distinct areas based on analysis of 10 brains. This atlas enables researchers to better understand brain functions and mechanisms of diseases.
The 2020 Joint Statistical Meetings feature cutting-edge research on COVID-19 modeling, including nowcasting and forecasting. Researchers also presented advancements in precision medicine for stem cell transplants, improving patient outcomes by extending lives.
Researchers at Cornell University have identified a problem that holds the key to whether all encryption can be broken. The study connects two areas of mathematics - cryptography and algorithmic information theory - to show that a natural computational problem characterizes the feasibility of basic cryptography.
Researchers at LMU Munich used smartphone data to analyze users' personalities, finding strong correlations between behavior patterns and self-assessed traits. The study focused on the Big Five personality dimensions, revealing that certain types of digital behavior are more informative for specific self-assessments.
Researchers at Ruhr-University Bochum developed a new method for efficiently identifying deep-fake images by analyzing objects in the frequency domain. The study found that images generated by GANs exhibit artefacts in the high-frequency range, which can be used to distinguish them from real photos.
Researchers develop new algorithm to measure crystal growth rate in supercooled liquids, achieving accuracy orders of magnitude higher than existing methods. The approach is based on molecular dynamics simulation and can be applied to various systems with different physicochemical properties.
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 at Carnegie Mellon University developed an efficient new way to quickly analyze complex geometric models by using Monte Carlo methods. This approach eliminates the need to divide shapes into meshes, reducing errors and increasing computation speed.
Researchers used recommender algorithms to suggest antiviral compounds based on latent relationships in chemical and biological data. They successfully pinpointed promising drug candidates with anti-SARS-CoV-2 activity.
Researchers at University of Cambridge developed a context-aware AI system that reduces the 'communication gap' by up to 96% for nonverbal people with motor disabilities, enabling faster conversation rates.
Researchers developed a framework using graph theory to optimize digital communication networks, finding cliques within them for efficient data transmission and improvement in throughput by up to 30%. This approach has far-reaching implications for future networks, including the internet of things with larger volumes of data.