Machine learning
Researchers from the University of Texas Austin have proposed a new model-constrained Tikhonov autoencoder neural network framework, called TAEN, capable of learning both forward and inverse surrogate models using a single arbitrary observational sample. This framework, TAEN, is designed to address the challenge of addressing scarce data regimes and overfitting
Machine learning
Researchers from the University of Texas Austin have proposed a new model-constrained Tikhonov autoencoder neural network framework, called TAEN, capable of learning both forward and inverse surrogate models using a single arbitrary observational sample. This framework, TAEN, is designed to address the challenge of addressing scarce data regimes and overfitting
Machine learning
Researchers at China Tobacco Sichuan Industrial Co. have successfully employed machine learning techniques in conjunction with FT-NIR spectroscopy to predict tobacco blend proportions with high accuracy. This innovative approach enables the development of accurate and reliable quality control systems, paving the way for improved cigarette quality and market reputation. The
Machine learning
A recent study from the New Jersey Institute of Technology has developed a competitive bidding framework to help contractors prepare for bundled contracts in transportation infrastructure. The framework involves machine-learning models that predict the level of competition on bundled contracts based on input variables such as contract characteristics. This research
Machine learning
A recent patent application filed by four inventors has described a system and method for providing alarm risk score analysis and intelligence in controlled environments. The system uses a combination of sensor information and video frames to determine the risk value of an event and generate an alarm if it
Machine learning
Researchers Prafulla Nath Dawadi, David Everton Norman, and Harel Moshe Yedidsion have filed a patent application for a method utilizing reinforcement learning to optimize substrate processing in facilities. The invention aims to maximize processing on higher-yield tools while meeting production thresholds. Key components include identifying current state data, providing it
Fatigue (Materials)
Research conducted by Argonne National Laboratory has presented a data-driven machine learning approach to modeling one-dimensional stress-strain behavior under cyclic loading, utilizing experimental data from the nickel-based Alloy 617. The approach employs uniaxial creep-fatigue test data acquired under various loading histories and compares two distinct neural network-based ODE models. Financial
Denial of service attacks
Researchers from the University of Milan have developed a new approach to detecting distributed denial-of-service (DDoS) attacks in Internet of Things (IoT) environments using federated learning with adaptive client selection. This method addresses the challenges of traditional centralized machine learning methods, which raise privacy and security concerns due to data
Information management
Researchers at Shanghai University have proposed a domain knowledge-assisted data anomaly detection (DKA-DAD) workflow to improve the accuracy of machine learning models in materials design and discovery. The study, funded by the National Natural Science Foundation of China and the National Key Research & Development Program of China, found that
Exports
A patent application by Jonathan Edward Hensley has outlined a method for the controlled release of information in data export. This approach aims to balance the need for access to sensitive information with the requirement for protection against unauthorized access. The proposal involves exporting data in a format with unmasked,
Machine learning
Researchers from Technology and Innovation Centre have proposed a new algorithm called Constrained Expectation Maximization (CEM) to address a limitation in teaching motion skills to robots through demonstrations. The CEM algorithm enforces time-sensitive constraints (TSC) during the learning process, enabling improved and more efficient reproduction of demonstration data. This breakthrough
Imperial College London
Researchers at Imperial College London have found that machine learning can accurately predict treatment response in patients with complex health conditions, but only when comprehensive clinical phenotyping is present. The study, published in PLOS One, simulated a clinical trial to compare the effectiveness of machine learning with traditional analysis in
