Articles tagged with Algorithms
A University of Washington team developed a highly capable five-fingered robot hand that can perform dexterous manipulation and learn from its own experience. The hand uses machine learning algorithms to model physics and plan actions, allowing it to adapt to new tasks without human intervention.
The algorithms enable a computer to act like a human when encountering an unknown word, looking up words in a dictionary and associating them with context. This development paves the way for computers to perform tasks such as translating texts, summarizing articles, and finding similar documents
A University of Michigan robot, MARLO, can now walk down steep slopes and uneven ground using a novel feedback control algorithm. The algorithm's potential benefits extend to powered prosthetic legs and other two-legged robots.
An experimental study found that readers prefer articles attributed to real journalists, but rate computer-generated texts as more credible. The study suggests that readers' expectations and limited experience with computer-based reports contribute to these differing assessments.
Researchers developed SpiroCall, a system that can accurately measure lung function using a standard phone call, regardless of smartphone availability. The tool's results were within 6.2% of clinical spirometer readings, making it suitable for monitoring chronic lung diseases at home.
Researchers have created a new machine-learning algorithm that reduces timing uncertainties in changing events by up to 300 times, allowing for more accurate dating and analysis of past events. The tool has applications in various fields, including geology, metrology, chemistry, biology, and astronomy.
A computer algorithm can recognize human emotions with high accuracy, transferring players' emotions to avatars or communicating with other players. The emotion detector system uses mathematical processing to analyze eyebrow position, eye openness, and mouth shape to correlate with basic human emotions.
A new decentralized planning algorithm for teams of robots factors in moving obstacles, reduces communication bandwidth, and preserves strong mathematical guarantees. The algorithm allows robots to operate in a world with static and dynamic obstacles while retaining their local and global objectives.
A team of researchers has identified new privacy concerns by demonstrating that geotagged posts on just two social media apps are enough to link accounts held by the same person. The study found that individuals can be identified with a high degree of confidence by matching their movements across two data sets.
Researchers at Southwest Jiaotong University developed an improved algorithm to estimate lithium ion phosphate battery state of charge by separately measuring charging and discharging states. This allows for more accurate estimation amidst initial inaccuracies and varying dynamic characteristics among batteries in series.
A novel approach, called machine unlearning, enables the removal of data without retraining a computer learning system from scratch. This method is crucial for increasing security and protecting user privacy, especially in the face of cyber-attacks and data breaches.
The Allen Institute will reconstruct over 1,000 brain connections using electron microscopes and machine learning algorithms. The project aims to advance artificial intelligence by reverse-engineering the brain's algorithms.
Researchers developed an algorithm that sorts and edits video footage into picturesque highlights for a vacation reel. The approach analyzes images with ideal artistic properties, considering factors like geolocation, composition, symmetry, and color vibrancy.
A new prediction tool developed by MIT engineers can detect rogue waves in calm waters, giving sailors a 2-3 minute warning. The algorithm analyzes wave groups and predicts the probability of a rogue wave developing based on their length and height.
Researchers at MIT developed a new model to reduce unnecessary communication in human-robot teams, enabling them to work together more effectively. The system has been tested on computer simulations and shows promise for improving team performance.
Researchers found that symmetrical RNA structures are harder to design, contradicting conventional thinking. The study uses data from over 100,000 players of the Eterna online game to develop a new rating scale for RNA-design difficulty.
A new planning algorithm generates backup plans efficiently and provides mathematical guarantees on its risk of failure, while also identifying conditions that trigger a switch to a particular contingency plan. The algorithm uses probabilistic pruning and optimistic heuristics to make the problem tractable.
Researchers developed AI software to teach a quadrocopter to autonomously recognize and follow forest trails. The drone was able to find the correct direction in 85% of cases, outperforming humans who guessed correctly 82% of the time. This breakthrough enables drones to complement rescue teams and accelerate searches for missing people.
A new algorithm combines traditional computer vision classification with deep learning models to improve pedestrian detection speed and accuracy. The technology has the potential to be used in smart vehicles, robotics, and image/video search systems.
Researchers will use insights from neural circuitry and learning methods to create more human-like computer vision and machine learning algorithms. The CMU-led team aims to unlock the brain's secret algorithms in learning and inference with a massive database of neural activity.
The Wyss Institute is leading a $21 million IARPA-funded brain mapping consortium to map neural circuits with unprecedented fidelity. The project aims to discover the brain's learning rules and synaptic circuit design, furthering neurally-derived machine learning algorithms.
Researchers develop a new algorithm to simulate electron trajectories in complex magnetic fields, significantly reducing simulation time. The method is applied to multipoles such as quadrupoles or sextupoles, yielding precise results and improving the stability of electron orbits.
A team of UMass Amherst computer scientists designed an algorithm to rapidly remove inventor ambiguity from patent records, improving searches. The winning approach will be incorporated into the USPTO's new online platform, PatentsView, providing users with more efficient and effective search capabilities.
Researchers apply shape optimization methods to enhance electric motor performance, achieving a 27% decrease in cost functional. The approach identifies optimal motor geometries that cannot be imagined beforehand, resulting in smoother rotation patterns and improved overall efficiency.
Researchers create a new method for generating training images to simulate natural variability, allowing for more accurate predictions and decision-making in various activities such as environmental cleanup and oil drilling. By analyzing a series of images using the demon algorithm, scientists select representative snapshots that captu...
Researchers developed a Bayesian Program Learning framework that captures human learning abilities, allowing computers to recognize and generate new visual concepts. The algorithm achieved impressive results in visual Turing tests, with only 25% of judges performing better than chance.
A new study leverages Statistics Canada data to identify prescription opioid-related deaths, providing a crucial gap in information for public health. The research algorithm shows high sensitivity and specificity, offering a valuable tool for researchers across Canada.
Researchers are studying how online dating technologies affect romantic connections, self-concepts, and decision-making processes. The project explores the social implications of updated online dating systems and their design features.
MIT researchers developed a new algorithm that reduces computation time for complex models by 200 times, using probability distributions and relevant data. The 'shrinking bull's-eye' algorithm can apply to various fields, including engineering, geophysics, and subsurface modeling.
Researchers developed a system that sends highly compressed images to servers, which then process the images and send back modified files with reduced bandwidth consumption. The technique reduces bandwidth usage by over 98% and energy consumption by up to 85%, making it ideal for mobile image processing applications.
Researchers at MIT have developed techniques to amplify minute motions in digital video, as well as smooth out small variations in digital images. The techniques can produce more polished images for graphic-design projects, while also disclosing structural defects and camouflaged objects.
A new study by Emilio Ferrara and Zeyao Yang found that positive emotions spread virally through Twitter feeds, with about 20% of users deemed highly susceptible to 'emotional contagion'. These users were four times more likely to produce positive tweets than negative ones.
Researchers developed NOMAD, a novel way to analyze bigger datasets using supercomputers and machine learning algorithms, achieving superlinear speedup on large-scale data. This breakthrough enables efficient processing of massive networks, predicting user preferences and identifying key topics in billions of documents.
An expert system can help diagnose schizophrenia's early onset symptoms, allowing for timely referrals and earlier treatment. The system uses AI to process reported symptoms and identify potential cases.
Researchers at Northeastern University have cracked the code of Uber's surge pricing algorithm, finding that prices can vary wildly based on location and demand. The team has developed a detailed map of surge areas, revealing that some areas offer significant discounts by walking or waiting five minutes.
Researchers have developed a new cutting-plane algorithm that improves upon existing methods, offering huge efficiency gains for specific optimization problems. The algorithm has been applied to various types of problems, including submodular minimization and semidefinite programming, resulting in significant speedups.
Researchers have presented new unconditional hardness results for dynamic and online problems, including matrix-vector multiplication and Patrascu's Multiphase Problem. These findings prove the existence of problems that are provably hard to solve, offering significant advancements in computer science.
Researchers have developed an algorithm that can analyze a person's intention despite interruptions or disturbances, allowing for more accurate control of robots and prosthetics. This technology could be used to guide cars on icy roads or assist stroke patients with mobility issues.
New algorithms have been developed using the natural process of Japanese tree frogs singing out of sync to solve computational problems in graphs and social networks. These algorithms, known as FrogCOL and FrogMIS, can identify independent sets of nodes and detect structural patterns in complex networks.
Researchers at MIT develop a new way of doing machine learning that enables semantically related concepts to reinforce each other. In experiments, their system outperformed conventional training methods in predicting Flickr tags and even when considering semantically similar tags.
The new algorithm enables filmmakers to determine the optimal number and location of cameras for capturing a scene, amassing data that can be used during post-production. It allows for virtual relighting, generating new viewpoints, and digitally recovering detailed 3D shapes and layouts.
The DOE is investing $10.5 million in CAMERA to develop computational mathematics and algorithms for experimental facilities research. This will help scientists extract useful information from experiments, tackling challenges such as data analysis and image processing.
A new algorithm using a highly sensitive troponin I assay can reduce mortality and cut triage times to one hour for patients with suspected acute myocardial infarction. The algorithm allows for safe discharge or rapid treatment initiation, enabling better resource utilization.
Researchers have created an algorithm that can efficiently search for the laws underlying complex dynamical systems, including biological processes. The algorithm, dubbed 'Sir Issac', was tested on artificial and real-world systems, demonstrating its ability to infer universal gravitational force and model chemical transformations.
A new study from NYU uses 7.7 million 311 service requests to track where and when neighborhood conflict arises, finding it more likely in areas with 'fuzzy' boundaries between homogenous communities. Conflict is less common in areas without clear boundaries or with clearly defined community groups.
A team of computer scientists from University of Utah discovered a way to find bias in algorithms used for hiring decisions and loan approvals. They developed a technique to determine if such software programs discriminate unintentionally, revealing potential issues with disparate impact. The researchers also found a method to fix thes...
Researchers developed an algorithm to analyze and critique digital photographs' composition, providing feedback on perceived aesthetics. The system relies on psychological theories of human vision and opinions of people, helping photographers improve their work.
The researchers developed a new algorithm using geometric matching and machine learning to analyze style similarity in 3D objects. They achieved an agreement rate of 90% with human evaluations, making it a valuable tool for computer graphics applications.
A new mathematical theory and algorithm, Multi-tiered iterative phasing (M-TIP), solves the reconstruction problem for fluctuation X-ray scattering data. This approach enables quick determination of general structure in minutes on a desktop computer, unlocking new advances in biophysics.
Researchers developed a 'reusable holdout' tool to identify meaningful correlations in adaptive data analysis. The algorithm, tested on two rigged datasets, correctly identified true statistical significance and eliminated false discoveries.
Researchers discovered that redeposition acts as a stabilising factor in the creation of hexagonally arranged dot patterns. The study reveals that eroded particles predominantly redeposit in valleys, not at hilltops, contributing to stable nanostructure formations.
A team led by Stanford electrical engineer Krishna Shenoy has developed a technique to make brain-controlled prostheses more precise, enabling people with spinal cord injuries to tap out commands with greater accuracy. The new approach continuously corrects brain readings to give users a more natural way to interact with devices.
Jennifer Tour Chayes of Microsoft receives the highest honor from SIAM, awarded for her groundbreaking work on phase transitions in mathematics and computing. She will deliver a prize lecture on massive networks on August 12.
Researchers developed an algorithm inspired by brain pruning, which eliminates unneeded connections in early childhood. The neuroscience-based approach created more efficient and robust networks than traditional engineering methods.
The book explores how hierarchical organization, wavelet transformation, subspace trees, and deep learning can overcome the curse of dimensionality to develop efficient big multimedia databases. It introduces essential statistical supervised machine learning algorithms for information retrieval.
A new learning method for energy optimization of plug-in hybrid electric buses uses stochastic Markov models to minimize fuel and electric consumption costs. The approach avoids the curse of dimensionality problem by approximating the cost function without discretization, reducing computational complexity.
Researchers developed a dropout-prediction model that uses data from one course offering to predict stopout in the next. The model achieved fairly accurate predictions and showed promise, particularly when incorporating additional variables like weekend study habits. Ongoing work aims to refine the model for improved accuracy.
A study using NASA satellite data found that collecting rainwater for vegetable irrigation in India can reduce water bills by up to 4,522 rupees per year. This can provide a profit of between 1,548 and 3,261 rupees per year, boosting cost savings and increasing quality of life.
Researchers at Hiroshima University found a correlation between telomere G-tail length and endothelial dysfunction, as well as stroke and dementia, in patients with chronic cerebrovascular disease. The study suggests that telomere G-tail length may be a useful biomarker for predicting cardiovascular risk.
Researchers at MIT report that a widely held assumption about computational complexity is correct, making it unlikely that a more efficient algorithm exists. The current edit distance algorithm, in use for over 40 years, has been deemed the best possible solution for comparing genomes or strings of symbols.