Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Muhammad Khalifa, a research scientist at NVIDIA, is advancing AI agents through his work on test-time techniques, reasoning agents, and LLM post-training. He completed his PhD at the University of Michigan in 2026 and p…
Databricks and Neo4j have partnered to create a next-generation digital twin that combines operational data with graph-based asset topology, enabling real-time fault propagation analysis and fleet-wide learning. The Airc…
INTERPOL's 2026 Global Financial Fraud Threat Assessment reports global fraud losses reached $442 billion in 2025, with financial fraud now among the top five global crime threats. A new graph-enriched Lakehouse approach…
Amazon Quick Research now enables researchers to integrate biomedical databases for rare cancer investigations by ingesting structured and unstructured data from sources like PubMed and applying LLM-driven synthesis to g…
A user on the Privacy Guides forum is seeking recommendations for private speech-to-text transcription tools. The discussion highlights Faster Whisper as a self-hosted option and mentions Lokas by Framasoft as a privacy-…
The CUDA caching allocator in PyTorch fragments memory when allocated blocks prevent adjacent free blocks from merging, causing allocation failures despite sufficient total free memory. This fragmentation occurs during C…
Amazon Web Services has extended support for the Model Context Protocol (MCP) in its Amazon Bedrock AgentCore Gateway, adding new capabilities for enterprise deployments including enhanced tool schema support, MCP prompt…
IP cameras record video for later review, while AI cameras analyze streams in real time and turn activity into structured events teams can act on. The shift from passive recording to active interpretation allows organiza…
General Motors Chief Product Officer Sterling Anderson said the automaker is entering a "third epoch" of engineering where artificial intelligence and machine learning are compressing development timelines from 15 hours …
Roboflow released a tutorial demonstrating how to combine its RF-DETR object detection model with Google's Gemini 2.5 Pro vision language model to automate soccer formation analysis from single broadcast images. The work…
OpenAI announced that one of its internal reasoning models solved a math problem first posed by mathematician Paul Erdős in 1946, concerning the maximum number of unit-distance pairs among points on a plane. The model di…
Stanford University's CS336 course has released official guidelines restricting how AI coding assistants may be used by students, prohibiting tools like ChatGPT and GitHub Copilot from writing code, completing assignment…
Tiling splits large images into smaller patches during computer vision inference, making small objects like cars and people in aerial imagery easier to detect by preserving detail that would otherwise be lost when the fu…
Amazon Web Services introduced AgentOps, an operational discipline for deploying and managing agentic AI at scale using Amazon Bedrock AgentCore. The framework addresses unique challenges of AI agents that make autonomou…
Amazon Web Services announced a new method combining Amazon FSx for Lustre with NVIDIA GPUDirect Storage (GDS) to reduce large language model cold-start loading times from minutes to seconds on AWS GPU instances. The app…
Amazon Quick has integrated Model Context Protocol (MCP) to allow financial analysts to query time-series databases using conversational language, eliminating the need for complex SQL queries. The integration connects Am…
Trustpilot built a real-time streaming pipeline using fine-tuned Google Gemma models to process millions of user reviews under strict latency and cost constraints. The company deployed a suite of specialized open-weight …
WindBorne Systems released WeatherMesh-6, an AI weather forecasting model that the startup claims surpasses the European Centre for Medium-Range Weather Forecasts (ECMWF) in accuracy and frequency, producing predictions …
PyTorch users can now implement custom operations in C++ and CUDA for use in both Python and C++ inference programs, with automatic device dispatch between CPU and CUDA implementations. The approach supports both statele…
LinkedIn re-architected its distributed linear programming solver, DuaLip, using a GPU-accelerated PyTorch version to solve extreme-scale optimization problems involving hundreds of millions of users and trillions of dec…