Explore the Newsverum news archive, editorial policies and categories in Turkish and English, with Arabic, French, Spanish, German and Portuguese editions. (Page 298 of 436)
An MIT-led machine-learning error-correction method improved forecasts over the two-to-six-week range, the researchers reported. The approach placed first in a European Centre for Medium-Range Weather Forecasts competition.
NVIDIA introduced the cuPotential library in early access and described semiconductor companies’ use of its CUDA-X tools for design, process modeling and inspection. The reported speedups are company claims from specific tests and applications.
IBM Research tested an agent factory using open-weight models and distributed inference to search hardware designs for a quantum error-correction decoder. Under the same 100-hypothesis search budget, the company reported inference costs of about $26 for the open-model configuration and $299 for the proprietary-model configuration.
IBM said it plans an early process design kit (PDK) so chip designers can explore its 0.7nm nanostack transistor architecture. The first version is expected in the first quarter of 2027 for research and development.
MIT researchers demonstrated an artificial neural network that can build its own learning machinery using simple connection-change rules. It learned pattern recognition and image classification, with performance reported as comparable to designed networks.
The Open Compute Project said IDC raised its forecast for annual spending on OCP-recognized IT infrastructure in 2029 to $800 billion. The earlier study forecast $295 billion for that year; the updated outlook reflects faster growth in infrastructure and AI investment.