Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
A new analysis explains how mode-agile threats render static library radar and electronic warfare systems ineffective, and how AI/ML cognitive architectures using artificial neural networks, deep neural networks, fuzzy l…
Feyn released FeyNoBg, an automatic background removal model, and open-sourced NoBg, a Python library for training and running such models. The model achieves the best published score on four of eight benchmarks and come…
Atlas Discovery launched out of stealth with foundation models that predict individual patient drug responses in clinical trials, targeting the costly gap between preclinical results and human efficacy. Y Combinator gene…
Whatnot, a live-shopping platform, presented a blueprint for observability at scale at Snowflake Summit 2026, detailing how it uses automated AI analysts and platform monitoring to handle billions of daily data points. T…
A new approach to test datasets for generalised machine learning tasks proposes drawing the test set from a distribution as different as possible from the training and validation sets, while still being in-distribution f…
A developer demonstrated the utility of Royston-Parmar (RP) models for causal inference in survival analysis, particularly in scenarios with non-proportional hazards where Cox proportional hazards models fail. The code, …
Booking.com relies on Arize AI for observability across its ecosystem of traditional ML models and generative AI agents, including trip planning assistants, ranking systems, and fraud detection models. The platform enabl…
A developer fixed a dangerous failure mode in vector search by making a single Supabase RPC the sole source of truth for search scope. The bug produced plausible-but-wrong neighbors from the wrong repository because the …
Spectro Cloud released a TCO calculator for its PaletteAI Inference Launchpad, a turnkey appliance that runs open models on local GPUs with intelligent routing, claiming it can cut inference costs by around 85% while kee…
Salk Institute neuroscientists have concluded that neural traveling waves in the visual cortex act as a computational engine, enabling the brain to build internal representations of the external world, according to a rev…
Microsoft has open sourced ML Video Codec (MLVC), a neural network-based video codec that reduces bandwidth to about one-eighth of H.264 while maintaining comparable quality, using Neural Processing Units (NPUs) for effi…
Mean Squared Error (MSE) loss does not incentivize neural networks to encode features in superposition, according to a new analysis by researchers Linda and Phil. The finding, demonstrated through mathematical proof and …
Researchers at an unnamed institution have developed a multimodal AI pipeline to analyze emergent non-verbal communication in the competitive video game Delta Force, marking a shift from prior studies focused on MOBA gam…
PIRAMID, a research initiative focused on statistical and mesoscopic theories of feature learning and generalization in neural networks, has released a progress update detailing team-by-team achievements and targets for …
University of Minnesota researchers have developed an AI system that enables underwater robots to track diver stress levels by analyzing exhaled bubbles, a breakthrough that could improve safety for commercial and scient…
A developer built a probabilistic graph neural network (PGNN) inference system for bio-inspired soft robotics maintenance, integrating ethical auditability from the start. The system models soft robots as graphs with pro…
A 2026 study (arXiv:2606.00060) using XGBoost on 70,000 hours of BTC/USDT data reports ~65% annualised return with Sharpe above 1 after applying a cost-aware execution filter, but the edge does not survive statistical si…
A researcher received an Honorable Mention in BlueDot's Technical AI Safety Puzzle #1 for training a small MLP to encode a country feature on a chosen nonlinear manifold in three reserved channels, demonstrating that a m…
A new open-source integration for Hugging Face Trainer, TraceML, diagnoses whether GPU underutilization is caused by data-loading bottlenecks or model compute limits. The tool, built by an unnamed developer, adds a singl…
A developer who grades AI models that write and judge code argues that the 'oracle problem' in software testing—determining whether an observed behavior is correct—remains unsolved despite being rebranded. Citing studies…