Articles tagged with Algorithms
Researchers at Oregon Health & Science University have developed an AI algorithm that uses data from continuous glucose monitors and wireless insulin pens to provide guidance on adjustments. The algorithm, paired with a smartphone app called DailyDose, generated recommendations in agreement with physicians 67.9% of the time.
A new study urges judges to scrutinize algorithmic risk assessment tools used in criminal cases due to biases and weak statistical evidence. The review recommends three steps: fitness, accuracy, and reliability assessments to ensure the tool's relevance, performance, and trustworthiness.
A team of engineers at the University of Cambridge has successfully trained a robot chef to prepare an omelette using machine learning, with improved taste and texture. The robot's culinary skills were refined to produce a reliable dish that meets human standards, overcoming the subjective nature of cooking.
The Fast Loaded Dice Roller (FLDR) algorithm simulates the roll of loaded dice to produce random integers with the best combination of speed, accuracy, and low memory requirements. FLDR can use up to 10,000 times less memory storage space than existing methods.
Researchers found that worker autonomy and transparency are crucial for crowdwork environment, while algorithmic management can enhance productivity when framed as promoting autonomy. The study suggests ways to improve algorithmic management via AI assistance.
Researchers developed an algorithm to predict coastal currents using dynamical systems theory and ocean data. This new tool promises to enhance search and rescue techniques at sea, potentially saving lives.
A new algorithm developed by researchers at the University of Córdoba improves response capacity to forest fires. The tool predicts fire fighting difficulty and offers an extinction difficulty index, helping authorities prioritize management tasks and optimize resources.
A new algorithm can continuously monitor a person's sympathetic nervous system activation using skin conductance and heart rate measurements, aiding in the development of wearable monitors for anxiety disorders.
A new study suggests that machine learning and AI in medical image reconstruction algorithms can result in myriad artefacts and major errors, leading to false positives and false negatives. The effects were typically not present in non-AI based imaging techniques.
Australian and US researchers used advanced computational methods to detect different structures of RNA, which until now could not be distinguished. The team discovered that changes in RNA structure influenced how the virus behaved, including which proteins it produced.
A recent study by Penn State researchers found that providing users with the option to adjust privacy and content settings on Amazon Alexa does not necessarily increase their trust in the platform. In fact, for power users, it can have an unfavorable effect. However, customizing content increased trust among sensitive users.
Fugu outperforms BBA in terms of least interruption time, highest image resolution and consistency of video quality, keeping viewers engaged for 5-9% longer
Researchers propose a new framework to quantify the predictability of temporal networks, which encodes the ordering and causality of interactions between nodes. The study found that the contributions of topology and temporality to network predictability vary significantly across different types of real networks.
A new startup, Oqullo, is developing a platform that utilizes computer visual algorithms and network cameras to gather demographic information about customers. This data can be used to offer personalized experiences in shopping, traveling, and other consumer activities.
Researchers at Ruhr-University Bochum used artificial intelligence to predict the structure of thin films, reducing the need for extensive experiments. The team developed a generative model that can generate images of the surface of a layer under specific process parameters, enabling the identification of optimal material formulas.
Researchers used a computer algorithm inspired by slime mould to create a three-dimensional map of the cosmic web structure in the local Universe. The analysis revealed that intergalactic gas is organised into filaments and detected at distances over 10 million light-years from galaxies.
Iowa State University engineers develop a closed-form solution for the inverse chirp z-transform, enabling efficient computation on the unit circle and lifting restrictions of existing algorithms. The algorithm achieves high accuracy with 64-bit floating-point numbers and operates in O(n log n) time.
Researchers at six universities, including UT Austin, are launching a partnership to turn AI 'black boxes' into human-understandable computer code. This will enable solving hitherto unsolvable problems in fields like drug design and cognitive science.
A new computer algorithm inspired by the mammalian olfactory system rapidly learns patterns and identifies smells even with strong sensory interference. The algorithm is applied to a neuromorphic computer chip, Loihi, which can learn to identify patterns or perform tasks a thousand times faster than traditional methods.
By using approximations instead of explicit kernels, researchers have accelerated machine learning speeds and improved AI's ability to handle large datasets. The new approach uses statistics to derive a nearly accurate kernel that can be computed much faster than traditional methods.
Researchers have successfully mapped the cosmic web's filamentary structure using a slime mold-inspired algorithm, providing insights into dark matter's role in shaping the universe. The study revealed that denser regions of intergalactic gas are organized into filaments that stretch over 10 million light-years from galaxies.
Researchers have developed an algorithm that uses the transit time of sound waves to assign echoes to specific walls. The drone's six degrees of freedom are sufficient for optimal microphone placement, reducing ghost wall detection. This innovation opens a new pathway towards practical applications in various fields.
Researchers developed a scalable method for estimating crop productivity in real time using satellite data, in-field camera networks, and ground measurements. The method provides highly accurate estimates of leaf area index (LAI) and can be used to detect underperforming fields or segments that need targeted management practices.
Researchers from UNSW Sydney have successfully analyzed the complex structure of benzene in 126 dimensions, shedding light on its stability and interactions. The discovery reveals unexpected electron behavior, where up-spin double-bonded electrons interact with down-spin single-bonded electrons.
A new algorithm has been discovered that can accurately register two point sets consisting of millions of points in a fraction of the time taken by conventional methods. The algorithm's accuracy is not compromised despite using approximations, and it has wide-ranging applications in computer graphics and computer vision.
Using Euclid's algorithm as an example, Dr. Lamport showed how a precise mathematical formula can be used to specify an algorithm, making it more efficient and easier to debug. By using math instead of code, developers can reduce the size of their programs by up to ten times and make debugging easier.
Computational biologists at Carnegie Mellon University have developed a new algorithm, MOCHI, to identify communities within the cell nucleus. The algorithm uses spatial arrangement and genetic interactions to subdivide interwoven nuclear components into groups.
A study by Stanford and UC Berkeley researchers found that algorithms are significantly more accurate than humans in predicting recidivism, approaching 90% accuracy. In contrast, human prediction rates were around 60%. The findings have important implications for criminal justice reform and the use of risk-assessment tools.
Researchers at Johannes Gutenberg University Mainz have developed an algorithm that can solve complex problems with remarkable facility on a personal computer. The SPA algorithm enables accurate weather forecasts with a prediction error of only 0.75 degrees Celsius, surpassing the performance of supercomputers.
Researchers at the University of Tokyo developed an algorithm that combines statistical modeling techniques with machine learning-based image recognition to analyze social media data. The algorithm predicts consumer purchases and allows brands to identify potential customers.
A team from Universitat Rovira i Virgili developed an algorithm that makes accurate predictions and generates interpretable models. This allows for a better understanding of the data, reducing biases in original data and providing valuable information for scientists.
University of Illinois researchers develop a new tire design that prioritizes puncture-free strength while maintaining elasticity for a shock-free ride. The study uses computer optimization to create unique structural patterns in the shear layer, ensuring the material can withstand pressure without failing.
A new AI algorithm developed by researchers can analyze retinal images to predict which patients with diabetic macular edema are likely to respond to anti-VEGF therapy. The algorithm achieved an 87% accuracy rate in predicting treatment response, offering a potential breakthrough in personalized medicine.
A Tel Aviv University study finds two writers composed 31 of 100 inscriptions from 8th-century BCE Samaria, providing insights into the logistical infrastructure of the kingdom of Israel and limited literacy rates in the early eighth century BCE.
Researchers have developed a custom artificial neural network that can analyze molecular signals controlling gene function, enabling biologists to understand complex mechanisms of gene regulation. This breakthrough enables the creation of machine learning algorithms that reflect common concepts in biology.
The MIT-developed model uses anonymized credit card transactions and quarterly reports to predict company earnings, outperforming human analysts on 57% of predictions. The researchers' method tackles the 'small data' problem by combining noisy signals with traditional financial data.
Researchers at Rensselaer Polytechnic Institute developed an algorithm to analyze biometric data from smart devices and recommend personalized sleep and light schedules. The approach has been shown to estimate circadian rhythm variation with high accuracy, potentially benefiting travelers and service members.
A new algorithm called OnTAD rapidly identifies and elucidates the functions of topologically associating domains (TADs), which are disrupted in diseases like cancer. The tool reveals increased gene expression in hierarchical TADs, associated with active epigenetic states.
A Wayne State University research team will develop algorithms to detect and prevent cyberattacks on chemical process control systems, enhancing production efficiency and safety. The three-year project aims to create stronger safeguards against automation system attacks.
A team of Brown University researchers used a brain-computer interface to reconstruct English words from neural signals recorded in the brains of nonhuman primates. The study showed that neural data produced high-fidelity reconstructions that were clear to a human listener, using multielectrode arrays for the first time.
A new blockchain algorithm called Proof-of-Search (PoS) has been proposed to reduce the environmental impact of Bitcoin mining by redirecting wasted energy towards solving optimization problems. The algorithm preserves the robustness of PoW while introducing miniblocks to prevent corruption.
A new study proposes using standard compression algorithms like zip software to calculate entropy, providing insights into protein folding and other complex systems. This method has endless applications in fields such as biomedical simulations and basic research.
Economists propose a novel approach to measuring inequality, accounting for individual relationships and social interactions. The new method correctly calculates the Gini coefficient by considering differences between individuals' wealth, not just individual values.
A new framework for designing machine learning algorithms helps prevent undesirable behavior, including biases in medical diagnoses and financial predictions. The approach shifts the burden from users to designers, making it easier to create 'well-behaved' algorithms without requiring extensive domain knowledge or complex data analysis.
Researchers at Stevens Institute of Technology have completed the first census of Wikipedia's 1,601 bots, shedding light on their functions and interactions with human users. The study found that bots play nine core roles, including fixers, connectors, and protectors, which together account for about 10% of all activity on the site.
Researchers at Stanford and UMass Amherst develop a new technique to create machine-learning algorithms that can learn to avoid undesirable outcomes such as gender bias and excessive risk. Their approach, called the Seldonian algorithm, enables users to specify what behaviors they want an AI system to avoid with high probability.
Researchers found that hiring algorithms are often opaque and biased, with few vendors disclosing concrete information on validation and mitigation. The study encourages transparency and conversation around ethical decision-making in pre-employment assessments using machine learning.
Researchers have created a spatial map of gene expression for individual cells in various tissues, including the liver and intestinal epithelium. The new algorithm, called 'novoSpaRc', uses machine learning to track gene activity and reveals new insights into tissue organization and regulation.
Researchers at Ludwig-Maximilians-Universität München have developed a new workflow for producing formaldehyde that increases yields by a factor of five, making it more efficient and sustainable than traditional methods. This breakthrough could significantly contribute to climate mitigation, with annual global production exceeding 20 m...
Researchers at Stevens Institute of Technology developed a meal-detection system that can accurately calculate carbohydrate intake in real-time, allowing for faster insulin administration. The system's estimates are remarkably accurate, outperforming previous systems by up to 17 grams.
A new co-regulation map of the human proteome has been created, enabling the prediction and assignment of functions to uncharacterized human proteins. The map reveals unexpected partnerships between proteins, including peroxisomal membrane protein PEX11β with mitochondrial respiration factors.
A team of international researchers has developed a new control algorithm that reduces the complexity of bilateral teleoperation systems while maintaining their performance. The proposed composited stated convergence scheme achieves this by reducing communication channels, resulting in better transient performance.
Researchers developed a compact depth sensor inspired by jumping spiders' impressive depth perception. The metalens sensor combines with an ultra-efficient algorithm to measure depth in a single shot, enabling applications in microrobots, small wearable devices, and virtual/augmented reality headsets.
A new study from UC Berkeley and Chicago Booth reveals significant racial bias in software programs that determine access to high-risk healthcare management programs. Fixing this bias could more than double the number of black patients automatically admitted to these programs.
Researchers developed an algorithm to recommend tags for social media posts, resulting in a 20% boost in popularity. The system takes into account user popularity and content emotional impressions to suggest effective tags.
A new study from the University of Illinois introduces a tool that helps quantify the connection between traffic accidents and city road networks. Researchers found that commuters willing to take longer trips can reduce accidents by up to 18%.
A team of researchers at KAUST has developed a mathematical bridge between high-level symbolic representations and statistical data, enabling artificial intelligence algorithms to make use of vast stores of symbolic knowledge. This breakthrough could lead to significant improvements in machine learning and biomedical applications.
Developed by Chung-Ang University researchers, the new exosuit helps users walk and run more efficiently while reducing energy expenditure. It features a fabric vest, wires, and batteries, weighing just 5kg and automatically switching between walking and running modes.
Researchers at MIT have created an algorithm that significantly speeds up the planning process required for robots to adjust their grasp on objects. The new approach uses motion cones to efficiently calculate feasible pushes and reposition objects in less than a second, compared to traditional algorithms that take over 500 seconds.
Researchers at Iowa State University have developed a new algorithm to solve the inverse chirp z-transform (ICZT), a problem that had gone unsolved for 50 years. The algorithm has been tested for numerical accuracy and shows promising results in terms of computational complexity.