Team streamlines neural networks to be more adept at computing on encrypted data
Research News Release
EurekAlert! provides eligible reporters with free access to embargoed and breaking news releases.
Eligibility GuidelinesEurekAlert! offers eligible public information officers paid access to a reliable news release distribution service.
Eligibility GuidelinesEurekAlert! is a service of the American Association for the Advancement of Science.
Researchers at the NYU Center for Cyber Security at the NYU Tandon School of Engineering are rethinking basic functions that drive the ability of neural networks to make inferences on encrypted data.
A collaboration across three continents at the frontiers of physics, biology, and engineering co-led by Maurizio Porfiri at NYU Tandon, applied super computing muscle and special software to a novel simulation of the Venus' flower basket sponge.
Researchers created a simulation of a deep-sea sponge and how it responds to and influences the flow of water. The work revealed a profound connection between the sponge's structure and function, shedding light on both the basket sponge's ability to withstand the dynamic forces of the surrounding ocean and its ability to create a vortex within the body cavity "basket." These properties may help for the design of ships, planes and skyscrapers of the future.
A new study of almost 12,000 Australians has found one-third of the adult population has experienced pure cybercrime during their lifetime, with 14% reporting this disruption to network systems in the past 12 months. With all forms of cybercrime already costing trillions every year globally, experts from the Australian Institute of Criminology (AIC) and Flinders University say the crimes involved substantial levels of personal victimisation including direct losses as well as the high cost of preventing future attacks.
Researchers are the first to model COVID-19 completion versus cessation in clinical trials using machine learning algorithms and ensemble learning. They collected 4,441 COVID-19 trials from ClinicalTrials.gov to build a testbed with 693 dimensional features created to represent each clinical trial. These computational methods can predict whether a COVID-19 clinical trial will be completed or terminated, withdrawn or suspended. Stakeholders can leverage the predictions to plan resources, reduce costs, and minimize the time of the clinical study.
For the first time an autonomously flying quadrotor has outperformed two human pilots in a drone race. The success is based on a novel algorithm that was developed by researchers of the University of Zurich. It calculates time-optimal trajectories that fully consider the drones' limitations.
Wearable devices can detect people's stress, according to new Washington State University research, opening potential new interventions for people with addictions. In a paper in the Journal of Medical Internet Research, a WSU research team found that wearable wristbands measure physiological responses to stress in real-time and real-world situations, providing a potential method to help people avoid slipping back into old behaviors.
A content recommendation system based on the user's brain model would be ideal for targeted advertising. Creating such a brain model, however, is computationally expensive. In a new study, researchers from Japan propose and validate a machine learning scheme to infer a user's brain model from their profile with high accuracy while optimizing the information collection cost using a feature selection technique, providing hope for its real-world application following further optimizations.
Although photovoltaic systems constitute a promising way of harnessing solar energy, power grid managers need to accurately predict their power output to schedule generation and maintenance operations efficiently. Scientists from Incheon National University, Korea, have developed a machine learning-based approach that can more accurately estimate the output of photovoltaic systems than similar algorithms, paving the way to a more sustainable society.
Scientists from the University of Graz, Kanzelhöhe Observatory, Skoltech, and the World Data Center SILSO at the Royal Observatory of Belgium, have presented the Catalogue of Hemispheric Sunspot Numbers. It will enable more accurate predictions of the solar cycle and space weather, which can affect human-made infrastructure both on Earth and in orbit.