Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers propose LieBN, a framework for Riemannian Batch Normalization over Lie groups that leverages left- and right-invariant metrics to normalize manifold-valued sample distributions. The framework is instantiated …
A machine-learning surrogate developed for the Modified Implicative Interdependency Model (MIIM) predicts contingency severity and component criticality in cyber-physical power networks, achieving Spearman correlations o…
A machine learning approach to automatic thematic indexing achieves F1 scores up to 0.67 on Voltaire's Complete Works, with the best-performing model from the Mistral family in a 4-bit quantised configuration, according …
Researchers propose StickyMoE, a differentiable routing consistency loss that reduces expert switch rates by up to 60% in Mixture-of-Experts models during pretraining, with less than 4% perplexity degradation, enabling m…
Researchers propose a unified approach to interpreting knowledge distillation (KD) for large language models (LLMs) by decomposing output scores into interactions, finding that KD methods work by sparsifying interactions…
A new study from arXiv (2607.09061v1) finds that vision models trained on global image inputs fail to generalize over task length or complexity due to reliance on global shortcuts, while recurrent vision policies based o…
A new study from arXiv introduces GenCeption, a model that leverages a pre-trained video generative diffusion backbone to perform general-purpose vision tasks, achieving state-of-the-art results in depth estimation, surf…
A new framework called DKCD (Domain Knowledge-enhanced Causal Discovery) improves causal discovery from unstructured data in high-expertise domains like healthcare, finance, and education. Developed by researchers, DKCD …
Researchers present ARCANA, a collaborative multi-agent framework for solving ARC-AGI-2 tasks under strict test-time and hardware constraints. The framework decomposes tasks into iterative perception, hypothesis generati…
Researchers at an undisclosed institution formalized a proof of well-posedness for the nonlinear Vlasov equation in the Lean 4 proof assistant using an AI-assisted strategy, completing the headline theorems in about a we…
A new dataset condensation method for action segmentation, Adaptive Latent Trajectory Anchoring, outperforms state-of-the-art techniques by shifting from optimization-based inversion to deterministic latent mapping using…
A new transformer architecture achieves 78% legibility in feed-forward operands and 50% in attention value channels by training operators to be crisp, bounded, and named as fuzzy set operations, researchers report in arX…
Vision Transformers (ViTs) robustly encode Gestalt-like figure-ground cues such as surroundedness and convexity from natural images, according to a study evaluating 25 ViTs with supervised and self-supervised training. R…
Researchers have developed SafeExplorer, an unbiased policy gradient modification for proximal policy optimization (PPO) that reduces training-time falls by up to 233x on physical robots while matching or exceeding final…
Leo, Technical Lead at Pangolinfo, presents a developer's guide to building an Amazon competitor research system using Python and the Pangolinfo API. The guide introduces the five-step IBADM framework and provides produc…
Sheaf, a new functional language for differentiable computation, brings Clojure's code-as-data paradigm to machine learning by representing models as inspectable, composable, and compiled data structures. The language el…
A developer built resolvo, an agentic pipeline that converts a GitHub issue into a tested pull request without requiring a local clone. Built with LangGraph and Google's Gemini models, the system routes issues by confide…
A new study explores how artificial intelligence is reshaping humanities research, enabling scholars in history, literature, and linguistics to analyze vast archives and uncover patterns that humans might miss. The integ…
A developer demonstrates how to build a resume-to-project matcher using fixed embedding vectors, showing the steps of normalization, cosine similarity scoring, and ranking. The example highlights that cosine similarity m…
An engineer's analysis reveals that LLM inference throughput on NVIDIA GPUs is bottlenecked by memory bandwidth, not compute. A 7B model achieves ~15 tok/s on a T4 but over 3,500 tok/s on an H100, a 150× gap that far exc…