Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Nokia's chief technology officer Pallavi Mahajan said the company's $1 billion investment from Nvidia will enable GPU-based radio access network (RAN) software that can run compute-intensive machine-learning algorithms l…
A developer has introduced the Data Trust Score Engine, an AI-powered cloud-native solution built on Google Cloud Platform that automatically measures and improves enterprise data reliability. The platform combines BigQu…
Anyscale co-founder Christian Stano outlines a three-step maturity curve for companies to build 'learning loops'—compounding flywheels of data curation, custom model training, and inference powered by proprietary data—as…
SkyRL now supports FP8-accelerated training and rollout with on-policy weight sync, reducing end-to-end step time by up to 23% while closely tracking BF16 convergence in long-run RL experiments. The system, developed by …
Anyscale and NVIDIA cuDF have integrated GPU-native operators into Ray Data, enabling up to 3x better total cost of ownership (TCO) on select data curation workloads compared to CPU-based solutions. The new features, ava…
Anyscale's Ray team reports that after a round of Ray Core improvements, batch inference on 500 nodes is 23% faster end to end, Ray Data shuffle is 24% faster, and placement groups are 62× faster to ready at 2,000 nodes …
A developer argues that coding agents fail repeatedly because their loops treat the context window as the only memory, discarding evidence from prior attempts. The developer proposes an append-only decision ledger stored…
In late September 2024, a nurse at Adventist Health in Bakersfield, Calif., stole fentanyl and morphine from a secured cabinet and used them herself while documenting administration to patients, according to investigator…
Google's inspect_viz library now supports heatmap generation for AI evals, enabling stakeholders to quickly compare model performance across skills. The tool, demonstrated with a script that wraps scores_heatmap(), colla…
SGLang and vLLM, the two major LLM serving engines, have attempted to enforce deterministic inference but face fundamental challenges due to floating-point non-associativity in GEMM kernels, which causes the same prompt …
Nvidia's DLSS 4.5 Ray Reconstruction (build 310.7.12) leaked ahead of its official rollout via the Call of Duty: Modern Warfare 4 PC beta and is now bundled in the DLSS Swapper tool, letting RTX owners enable the second-…
Uniform-Cost Search (UCS) is introduced as a graph traversal algorithm that finds the cheapest path by considering accumulated path cost, unlike DFS or BFS. In a weighted graph example with edge costs from START to TREAS…
ShrinkRay, a new open-source command-line tool, compresses and quantizes small neural networks for microcontrollers, providing a fit verdict against a 12-chip database before firmware builds. The tool, installable via `p…
AI-driven trading is rewriting the rules of profit season, with machines listening to earnings calls in real time and trading on nuances before humans can react, causing sharp share-price swings. Commonwealth Bank's shar…
Vladimir Besk released the complete personal archive of Konstantin Tsiolkovsky as a dataset on Hugging Face, comprising 51,008 sheets across 2,019 archival files under CC0. The archive includes 1,759 manuscript-typescrip…
LinkedIn's principal AI researcher Praveen Bodigutla said the company built a cognitive memory agent to provide agentic personalization for its hiring assistant, moving off GraphRAG in favor of a tree-structured memory f…
Researchers introduced EXPL-FR, a method that explains face recognition (FR) models by aligning a vision-language model's (VLM) image encoder with the frozen FR embedding space, enabling attribute-level explanations with…
SchemaRouter, a lightweight routing layer for heterogeneous agentic retrieval-augmented generation (RAG) systems, achieves answer accuracy of 0.71 on a materials-science benchmark of 110 queries while using 227 retrieved…
A new arXiv study (2608.21408v1) comparing hate speech classification techniques for Roman Urdu found that parameter-efficient fine-tuning (PEFT) with LoRA and prompt engineering methods offer computationally efficient s…
A systematic review of software frameworks for explainable AI (XAI) in time series classification (TSC) finds that only six frameworks explicitly support time series, with just one offering frequency-domain explanations …