Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Researchers have introduced state commitment learning, a new training objective that teaches language models to distinguish temporary computation from persistent memory. The method, called Counterfactual Erasure RL (CERL…
Researchers have developed ReasoningFlow, a framework that maps the non-linear discourse structures of large reasoning model (LRM) traces into directed acyclic graphs (DAGs) to improve evaluation and monitoring. Through …
A new stereological theory reveals that current LLM benchmarks suffer from a structural blind spot exceeding the runner-up score gap by two orders of magnitude and dominating statistical noise by 52-127x. Analysis of thr…
Researchers introduced a two-stage early failure alerting system for dialogs and LLM-agent trajectories that learns from sparse turn-level evidence rather than assigning terminal labels to every prefix. The attention-bas…
Researchers introduced trajectory extrapolation error, a measure of how much a language model's hidden states deviate from a linear path during word processing, and found it independently predicts human reading times bey…
Researchers introduced VideoKR, a large-scale training corpus of 315,000 video reasoning examples over 145,000 expert-domain videos designed to strengthen knowledge- and reasoning-intensive video understanding. The team …
Researchers have developed camroll-agent, a conversational AI assistant that accesses a user's personal camera roll to answer questions about their photos, from factual queries like identifying food eaten yesterday to op…
Researchers have developed LightVesselNet, a neural network with only 75,000 parameters that performs retinal blood vessel segmentation competitively with much larger models. The ultra-lightweight architecture, which use…
Researchers have introduced MCBench, a new benchmark designed to assess the safety of Omni Large Language Models that process vision, audio, and text. The benchmark includes 1,196 scenarios across four safety categories,…
Researchers have developed RePHO, a method that reconstructs physically plausible human-object interactions from monocular videos by combining kinematic estimates with reinforcement learning. The approach addresses commo…
Researchers released Biomazon, a 20-meter multimodal benchmark dataset covering the Amazon Basin that pairs GEDI relative height (RH) profiles and aboveground biomass density (AGBD) targets with satellite and land cover …
Researchers have developed LANTERN, a lightweight memory layer that recovers facts lost when large language models compress long conversations, achieving 78.3% recovery of verifiable facts without requiring any LLM calls…
Researchers have developed a data-efficient flood depth prediction method that uses domain-aware coreset selection and tabular foundation models, achieving near-real-time accuracy with only 0.7% of the training data typi…
Researchers have developed TTT-RTL, the first per-design test-time training framework that uses reinforcement learning to optimize hardware designs generated by large language models. The system closes the loop between a…
Researchers have developed the Multi-Granularity Reasoning Network (MGRN), a new framework for natural language inference that explicitly leverages hierarchical semantic features across multiple levels of language unders…
Researchers developed and evaluated deep learning models for automated age-related macular degeneration (AMD) staging using OCT and OCT angiography data from 271 participants. A biomarker-based model achieved the highest…
Researchers developed a phase-aware large language model agent that optimizes retrieval system parameters by conditioning each proposal on its full optimization history, overcoming the limitations of traditional hyperpar…
Researchers have developed DFPL, a framework for incomplete image-tabular classification that addresses the missing-modality problem through fine-grained prototype learning. The method introduces Shared-Specific Prototyp…
A new study shows that discrete Gradient Descent (GD) with a large step size restores symmetry in multi-pathway Deep Linear Networks, counteracting the "winner-takes-all" specialization predicted by Gradient Flow. Resear…
Researchers have demonstrated that formal concept lattices, derived from Formal Concept Analysis, can serve as principled semantic scaffolds to guide neural network learning by organizing concepts hierarchically from gen…