Articles tagged with Artificial Intelligence
A team of researchers at USC has developed a system to quantify and fix anti-queer bias in AI-powered text prediction. By training the popular language model BERT on queer-inclusive data from Twitter and news articles, they significantly reduced biases in predictions, with QueerTwitter showing the most effective results.
MIT researchers developed a method to create 3D-printed materials with tunable mechanical properties and embedded sensors, enabling real-time feedback on movement and interaction. The sensing structures use air-filled channels that deform when moved or squeezed, providing accurate feedback for robotics and wearable devices.
Researchers developed a Flashover Prediction Neural Network (FlashNet) model to forecast deadly fire events, beating other AI-based tools with up to 92.1% accuracy across various building floorplans. The model's performance improved when given real-world data, highlighting its potential for saving firefighter lives.
Researchers at UNH will develop and test social assistive robots to aid in the care of individuals with Alzheimer’s disease and related dementia in their own homes. The robots will be equipped with smart home devices, artificial intelligence, and wearables to track patients' activity levels and health management.
A new study by Petra Nieken found that charismatic leadership tactics and consistent video communication lead to better performance in remote work. In contrast, traditional charisma questionnaires did not predict staff performance. The study suggests that managers should convey a consistent impression when using the video channel.
Researchers have developed a proof-of-concept model that uses artificial intelligence to combine multiple types of data from different sources to predict patient outcomes. The models demonstrate the ability to make prognostic determinations while uncovering the predictive bases of features used to predict patient risk.
Researchers at Carnegie Mellon University have developed an AI pilot that can safely navigate crowded airspace by predicting the intent of other aircraft and avoiding collisions. The AI uses vision and natural language processing to communicate with other aircraft, enabling safe and socially compliant navigation.
Researchers developed a flexible, stretchable computing chip that processes information like a human brain to analyze health data. The device aims to change the way health data is processed, enabling continuous tracking of health without sending data wirelessly.
The University of Illinois School is awarding a grant to explore the use of conversational AI in libraries, aiming to strengthen engagement between libraries and diverse audiences. The project will investigate ways to democratize conversational AI for librarians, enhancing workforce development and community service.
Researchers at Washington University in St. Louis developed a new imaging tool that uses optical coherence tomography (OCT) and machine learning to detect and classify cancerous tissue samples with high accuracy. The technique, which was tested on patients in a trial, showed a 93% diagnostic accuracy rate.
A new study used artificial intelligence to create the largest global map of insect diversity, highlighting areas with high ant biodiversity and potential conservation priorities. The researchers also found that only a low percentage of these areas are protected.
Physicists have created a way to simulate quantum entanglement between interacting particles using neural networks and fictitious 'ghost' electrons. This approach enables accurate predictions of molecule behavior, which could lead to breakthroughs in pharmaceutical development and material design.
Researchers at MIT have developed a machine-learning system that uses computer vision to monitor the 3D printing process and correct errors in real-time. The system successfully printed objects more accurately than other 3D printing controllers, enabling engineers to incorporate novel materials into their prints with ease.
A new study introduces a novel epigenetic predictor, PCBrainAge, that captures aging heterogeneity across multiple brain regions. The tool demonstrates stronger associations with AD dementia and pathologic AD compared to existing age predictors.
Researchers evaluated the real-time performance of an AI model to detect COVID-19 from chest X-rays and found that it did not outperform radiologist predictions. The study highlights the limitations of AI in diagnostic accuracy, particularly for certain groups such as women and minority populations.
A new AI-based dynamic brain imaging technology has been introduced by Carnegie Mellon University, which can map out rapidly changing electrical activity in the brain with high precision and speed. The technology uses deep learning approaches to translate scalp EEG signals back to neural circuit activity without human intervention.
MIT researchers have developed a new type of programmable resistor that enables analog deep learning, which promises faster computation with reduced energy usage. The device can process complex AI tasks like image recognition and natural language processing, paving the way for integration into commercial computing hardware.
A Nagoya University research group has developed an AI algorithm that can diagnose idiopathic pulmonary fibrosis with high accuracy, based on non-invasive examinations and medical data. The technology may revolutionize medical care by allowing doctors to request AI-assisted diagnoses instead of specialists.
A study published in Science Robotics journal reveals that humans are unable to distinguish between a humanoid robot's response time variability and that of a human when interacting with the robot. This finding suggests robots can exhibit human-like behavior, raising questions about the blurring line between humans and machines.
A new study found that Black patients are six times more likely to have advanced vision loss after a glaucoma diagnosis, highlighting the need for early eye screening in this population. The research analyzed nearly 210,000 participants and used artificial intelligence to identify visual field loss patterns.
Researchers at USC's Information Sciences Institute developed a method to train AI to understand analogies in Aesop's fables, enabling it to make creative connections between familiar and novel situations. The study found that humans approach analogical reasoning subjectively and interpretively, influencing the outcome.
Researchers have developed a novel simulation method to calculate free energy using deep learning artificial intelligence, providing accurate models of molten salts' thermodynamic properties. The study could help examine corrosion in metal containers and improve the design of next-generation nuclear reactors.
Researchers developed an AI-driven device that detects and predicts hemodynamic instability using a single ECG lead, outperforming traditional vital sign measurements with nearly 97% sensitivity. The technology has the potential to provide continuous dynamic monitoring capabilities in patients with intermittent static vital sign measur...
A new study uses machine learning models to predict cancer patients' responses to immunotherapy based on their gut microbiome features. The research identifies common gut bacterial taxa associated with responders versus non-responders, providing a potential tool for distinguishing and predicting immunotherapy responders.
The University of Hong Kong's Faculty of Dentistry has developed an AI-powered approach to automate denture design, allowing for high accuracy and reducing treatment time. The new method uses a 3D Generative Adversarial Network (3D-GAN) algorithm to reconstruct the shape of natural teeth and generate prosthetic teeth.
A University of Ottawa-led team has developed an AI-based deep learning model to identify cystic hygroma, a rare and life-threatening disorder, from first-trimester ultrasound scans with high sensitivity and specificity. The approach may be applied to other fetal anomalies identified by ultrasonography.
Developed a dataset of over 2,000 clinically relevant questions written by medical experts to train AI models. The model asked high-quality and authentic questions over 60% of the time, compared to real questions from medical experts.
Researchers from Johannes Gutenberg University Mainz used AlphaFold to predict the structures of new protein knots, discovering the most complex knot and composite knots. These findings provide insight into folding mechanisms and evolutionary processes in proteins.
Researchers developed an AI system that classifies IBDN lesions accurately, displaying image-based diagnostic ability with 64.5% sensitivity and 89.5% specificity. The correct diagnosis rate of the AI system was 79.0, surpassing that of endoscopists, who achieved a 77.8% accuracy rate.
Researchers discovered that search results yield male-dominated output despite neutral searches, promoting gender bias and potentially affecting hiring decisions. Exposure to such outputs influences perceptions and decisions, reinforcing societal inequalities.
Researchers from North Carolina State University have developed a new approach to federated learning that allows them to develop accurate AI models more quickly and accurately. By reformating updates sent to the centralized server, they can resolve the heterogeneity problem in data, improving model performance.
Researchers from Osaka University found that facial similarity plays a crucial role in ratings of trustworthiness for observers of the same sex, but not for observers of opposite sex. The study suggests that facial similarity is an important factor affecting social judgments for same-sex interactions.
A study by University of Sydney neuroscientists found that people's brains can detect AI-generated fake faces with a 54% accuracy rate. However, verbal identification was only successful 37% of the time, highlighting the potential for using brain activity to flag deepfakes on digital platforms.
Researchers found that mere exposure to a human-like robot is not enough to make people believe it is capable of thoughts and emotions. Human-like behavior is crucial for being perceived as an intentional agent. The study aims to inform the design of social robots in the future.
Researchers developed a convolutional neural network to identify fetuses with Down Syndrome from ultrasound images. The model achieved high accuracy, improving detection by over 15% compared to existing methods. Non-invasive screening could become a convenient and inexpensive tool for early pregnancy diagnosis.
Researchers at Saarland University are developing orthopaedic implants that can actively stimulate bone healing by undergoing controlled micromotions. These innovative implants, which use shape-memory wires, can provide real-time information on fracture progression and identify potential complications.
Researchers developed a neural network algorithm that recognizes emotions and engagement from video images of faces, outperforming existing models in accuracy. The system can be integrated into video conferencing tools and online learning systems to analyze participant engagement and emotions.
A new theory developed by a collaboration between a former cosmologist and a computational neuroscientist has identified essential connections between brain cells. The theory, published in Physical Review Research, uses geometric framework to predict structure from function in neural networks.
Hawks use a unique flight path to slow down to a safe speed while minimizing the distance from the perch at which they stall. This allows them to control their landings effectively, even when slowing down risks stall, leading to sudden loss of flight control.
A team of researchers led by Danilo Vasconcellos Vargas has developed a new method called 'Raw Zero-Shot' to evaluate the robustness of artificial neural networks in image recognition. The study found that Capsule Networks produced the densest clusters, indicating improved transferability and potential solutions for improving AI robust...
By using the brain's visual response as feedback, researchers can reconstruct images of simple objects in real-time. The technique has potential applications in augmenting human capabilities and could one day be used to bring together human and artificial intelligence.
A new robotic system, FuseBot, has been developed to efficiently retrieve buried objects in piles. The system uses radio frequency signals and computer vision to reason about the probable location and orientation of objects under the pile, enabling it to find more hidden items than a state-of-the-art robotics system in half the time.
A team of experts presents a new framework for understanding the complex relationship between AI and climate change, suggesting ways to better align AI with climate goals. The study highlights the need for holistic evaluation of AI's impact on greenhouse gas emissions.
A joint research group developed a robotic AI system that autonomously determines optimal conditions for growing replacement retina layers. The system achieved a 90% rate of differentiation efficiency in 185 days, compared to months of work by humans.
Researchers found that two types of computer models, deep neural networks and the Basel Face Model, are surprisingly accurate at replicating human facial similarity judgments. These models were tested on a dataset of realistic computer-generated faces and ranked by students in terms of similarity.
Three African vulture species have significantly larger home ranges than existing protected areas, increasing exposure to human-related food sources and environmental toxins. Conservation efforts require addressing challenges outside protected areas.
Researchers found that people are less morally outraged when gender discrimination occurs due to an algorithm rather than direct human involvement. The study's findings have broader implications for efforts to combat discrimination and may affect how companies are held liable.
A new deep learning model can accurately identify disease-related features from images of eyes, even in low-quality images. The lightweight model is deployable on mobile devices and has better segmentation accuracy than other commercial software.
Researchers at the University of Tokyo have developed a new method to detect deepfakes, using self-blended images that improve detection accuracy by 5-12%. The team created novel synthesized images with controlled artifacts to train algorithms and found significant improvements in detecting deepfake images and videos.
A Rutgers team created an AI-powered framework that detects trespassing events, differentiates violators, and generates video clips of infractions. The system analyzed 1,632 hours of archival footage and found nearly 70% of trespassers were men, with most incidents occurring on Saturdays around 5 p.m.
Researchers at NIST developed LANTERN, a new statistical tool that can predict protein function while providing interpretable explanations. The tool has shown promise in predicting genetic edits for COVID-19, E. coli, and GFP proteins, achieving high predictive accuracy.
A new AI system uses artificial neural networks to recognize objects more accurately and stably, despite changing visual inputs. The system mimics human eye movements to improve machine vision capabilities, reducing errors in self-driving cars and other applications.
A team of researchers from Kyushu University has developed an olfactory sensor capable of identifying individuals by analyzing the compounds in their breath. The system, combined with machine learning, achieved an average accuracy of over 97% in authenticating up to 20 individuals.
A new optimization tool can quickly improve performance of various autonomous systems, including walking robots and self-driving vehicles. The tool uses automatic differentiation to identify tweaks that achieve desired outcomes, reducing trial-and-error simulations.
Researchers at MIT created insect-scale robots that can emit light during flight, allowing for precise motion tracking and potential communication between robots. The ability to emit light also enables the robots to call for help in search-and-rescue missions.
Researchers at the University of Rochester have discovered a novel neural mechanism involved in causal inference that helps the brain detect object motion during self-motion. This discovery may have applications in designing artificial intelligence devices and developing treatments for brain disorders such as autism and schizophrenia.
Computer scientists have developed a system to analyse noise produced by individual cameras, linking videos or images to specific cameras. The tool extracts camera-specific noise from video frames, achieving high accuracy rates and potential practical use for law enforcement.
A team of researchers developed a framework using artificial intelligence to improve long-term life satisfaction. The model, called a self-organizing map, estimates an individual's psychological age, future well-being, and risk of depression, providing personalized daily tips for improved well-being.
Researchers developed an AI method to automatically detect plaque erosion in heart arteries using OCT images. The new technique uses neural networks and post-processing algorithms to predict regions of possible plaque erosion and refine the initial prediction based on clinically interpretable features.
Researchers use machine learning to automatically analyze Reflection High-Energy Electron Diffraction (RHEED) data, enabling faster and more efficient discovery of new materials. The study focused on surface superstructures in thin-film silicon surfaces and identified optimal synthesis conditions using non-negative matrix factorization.