Loading…

AI & Machine Learning News Hub

Research, releases, and applied work in AI & ML

What's New

Top 5 Across All Sources

  1. New chip could help tiny robots traverse complex environments

    MIT News - Artificial intelligence · 6h ago
Latest
MIT News - Artificial intelligenceNew chip could help tiny robots traverse complex environmentsMIT News - Machine learningNew chip could help tiny robots traverse complex environmentscs.LG updates on arXiv.orgTowards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6Gcs.LG updates on arXiv.orgNeuroShield: A Device-Agnostic Foundation Model for EEG Authenticationcs.LG updates on arXiv.orgMassive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Testcs.LG updates on arXiv.orgCIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networkscs.LG updates on arXiv.orgEvidential Fusion Network for Multimodal Survival Prediction under Missing Modalitiescs.LG updates on arXiv.orgELADO: Elliptic PDE Assessment Datasets for Operator Learningcs.LG updates on arXiv.orgB[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNetcs.LG updates on arXiv.orgCELEUS: Certifiable and Efficient LLM Evaluation via E-Processescs.LG updates on arXiv.orgEvolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learningcs.LG updates on arXiv.orgMachine Learning Classification of Cryopathy Syndromes: A Comprehensive Comparative Studycs.LG updates on arXiv.orgUnderstanding Latent Flow Models for Tabular Data Synthesis: Targets, Paths, and Samplingcs.LG updates on arXiv.orgTemporal Causal Prior-Data Fitted Networks for Panel Data with Learned Reliability Signalscs.LG updates on arXiv.orgMMGNN: Multi-level, multi-color graph neural networks for molecular property predictioncs.LG updates on arXiv.orgPhysics-Guided Dual-Stream Heterogeneous Graph Neural Network for Predicting Full-Field Structural Response of Stiffened Panelscs.LG updates on arXiv.orgShort-Term Electricity Demand Forecasting for New England Using a Hybrid Transformer-XGBoost Framework with Weather, Calendar, and COVID-19 Indicatorscs.LG updates on arXiv.org$\Omega$: Operator-based Mixture Ensemble for Generative Assimilationcs.LG updates on arXiv.orgHierarchical Pooling for Sheaf Neural Networkscs.LG updates on arXiv.orgTowards Robust Training in NNGPT AutoML Pipeline: A Loss-Optimizer Pairing Selection StudyMIT News - Artificial intelligenceNew chip could help tiny robots traverse complex environmentsMIT News - Machine learningNew chip could help tiny robots traverse complex environmentscs.LG updates on arXiv.orgTowards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6Gcs.LG updates on arXiv.orgNeuroShield: A Device-Agnostic Foundation Model for EEG Authenticationcs.LG updates on arXiv.orgMassive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Testcs.LG updates on arXiv.orgCIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networkscs.LG updates on arXiv.orgEvidential Fusion Network for Multimodal Survival Prediction under Missing Modalitiescs.LG updates on arXiv.orgELADO: Elliptic PDE Assessment Datasets for Operator Learningcs.LG updates on arXiv.orgB[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNetcs.LG updates on arXiv.orgCELEUS: Certifiable and Efficient LLM Evaluation via E-Processescs.LG updates on arXiv.orgEvolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learningcs.LG updates on arXiv.orgMachine Learning Classification of Cryopathy Syndromes: A Comprehensive Comparative Studycs.LG updates on arXiv.orgUnderstanding Latent Flow Models for Tabular Data Synthesis: Targets, Paths, and Samplingcs.LG updates on arXiv.orgTemporal Causal Prior-Data Fitted Networks for Panel Data with Learned Reliability Signalscs.LG updates on arXiv.orgMMGNN: Multi-level, multi-color graph neural networks for molecular property predictioncs.LG updates on arXiv.orgPhysics-Guided Dual-Stream Heterogeneous Graph Neural Network for Predicting Full-Field Structural Response of Stiffened Panelscs.LG updates on arXiv.orgShort-Term Electricity Demand Forecasting for New England Using a Hybrid Transformer-XGBoost Framework with Weather, Calendar, and COVID-19 Indicatorscs.LG updates on arXiv.org$\Omega$: Operator-based Mixture Ensemble for Generative Assimilationcs.LG updates on arXiv.orgHierarchical Pooling for Sheaf Neural Networkscs.LG updates on arXiv.orgTowards Robust Training in NNGPT AutoML Pipeline: A Loss-Optimizer Pairing Selection Study

By Source

Feeds organized so you can skim by site.

Density Sort
MN
New chip could help tiny robots traverse complex environments 6h ago Gleanmer is a new system that can construct detailed 3D maps of a robot’s environment at high speed while operating at extremely low power. The advance could enable tiny devices to avoid obstacles and safely navigate in the real world. A better way to model the behavior of metal alloys 3d ago MIT researchers created a technique that captures chemical arrangements across materials to improve predictions of how metal alloys and other complex materials will behave. MIT in the media: For the future of tech, "Massachusetts can absolutely lead" 5d ago Leaders, faculty across MIT discuss fostering innovation and talent in Greater Boston in special series of articles published alongside the outlet's annual list of 'Tech Power Players' In game theory, generalists sometimes win out over specialists 5d ago In a new paper, MIT LIDS researchers show that for certain kinds of games, an overlooked class of algorithms performs much better than expected against better-trained opponents. Could AI tell you where you left your keys? 6d ago A new memory framework known as DAAAM enables a robot to rapidly recall rich descriptions and precise locational information about objects it encountered while exploring its environment. This efficient approach could help an autonomous agen... MIT’s Initiative for New Manufacturing builds momentum 6d ago MIT’s Initiative for New Manufacturing (INM) is advancing research, workforce development, and industry collaboration to accelerate new manufacturing technologies, strengthen industrial competitiveness, and help shape the future of manufact... Jinhua Zhao named head of the Department of Urban Studies and Planning 11d ago MIT Professor Jinhua Zhao, a noted scholar and transportation planner, has been appointed head of the MIT Department of Urban Studies and Planning. When it comes to predicting people’s preferences, it pays to consider “the power of three” 11d ago An MIT team proved that it is impossible to get information about correlations from two-way comparisons alone. Correlations can be discerned, however, when large numbers of people rate three alternatives in their order of preference. MIT affiliates win 2026 Hertz Foundation Fellowships 11d ago Three MIT students and an incoming graduate student have won 2026 Hertz Foundation Fellowships. The fellowships in applied sciences, engineering, and mathematics recognize doctoral students who are pursuing solutions to the most pressing ch... Startup’s nuclear-inspired cooling system could make data centers more sustainable 13d ago The startup Ferveret, founded by MIT Associate Professor Matteo Bucci and former MIT postdoc Reza Azizian, reduces the energy required to cool chips in data centers that power AI.
20 loaded
MN
MIT News - Machine learning
6h ago · 20 items
New chip could help tiny robots traverse complex environments 6h ago Gleanmer is a new system that can construct detailed 3D maps of a robot’s environment at high speed while operating at extremely low power. The advance could enable tiny devices to avoid obstacles and safely navigate in the real world. A better way to model the behavior of metal alloys 3d ago MIT researchers created a technique that captures chemical arrangements across materials to improve predictions of how metal alloys and other complex materials will behave. In game theory, generalists sometimes win out over specialists 5d ago In a new paper, MIT LIDS researchers show that for certain kinds of games, an overlooked class of algorithms performs much better than expected against better-trained opponents. Could AI tell you where you left your keys? 6d ago A new memory framework known as DAAAM enables a robot to rapidly recall rich descriptions and precise locational information about objects it encountered while exploring its environment. This efficient approach could help an autonomous agen... When it comes to predicting people’s preferences, it pays to consider “the power of three” 11d ago An MIT team proved that it is impossible to get information about correlations from two-way comparisons alone. Correlations can be discerned, however, when large numbers of people rate three alternatives in their order of preference. Startup helps retailers track their products in real-time 18d ago The startup Cartesian is helping retailers keep track of inventory with a technology invented at MIT. Using wireless signals from attached RFID tags, the system finds items’ precise location in a store, from the stockroom to the shop floor. NSF renews support for MIT-led AI and physics institute, expanding a new model for discovery 18d ago The MIT-led Institute for Artificial Intelligence and Fundamental Interactions (IAIFI) has received renewed support from the National Science Foundation for an additional five years. Teaching AI agents to ask better questions by playing “Battleship” 19d ago AI models played “Collaborative Battleship” together and struggled to ask informative questions about hidden ships. A Monte Carlo inference strategy helped small agents carefully consider each inquiry to outperform larger systems at a fract... MIT researchers teach AI models to interpret charts 20d ago Researchers used a novel data generation pipeline to build ChartNet, a large synthetic dataset of chart images paired with corresponding information. They used this training dataset to improve the performance of generative AI models at chal... MIT economist Whitney Newey awarded Erwin Plein Nemmers Prize in Economics 32d ago MIT economist Whitney Newey was awarded the Erwin Plein Nemmers Prize in Economics, a biennial award from Northwestern University that celebrates lasting contributions to economic scholarship.
20 loaded
CL
cs.LG updates on arXiv.org
6h ago · 20 items
Towards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6G 6h ago Abstract page for arXiv paper 2606.20670: Towards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6G NeuroShield: A Device-Agnostic Foundation Model for EEG Authentication 6h ago Abstract page for arXiv paper 2606.20673: NeuroShield: A Device-Agnostic Foundation Model for EEG Authentication Massive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Test 6h ago Abstract page for arXiv paper 2606.20743: Massive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Test CIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networks 6h ago Abstract page for arXiv paper 2606.20747: CIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networks Evidential Fusion Network for Multimodal Survival Prediction under Missing Modalities 6h ago Abstract page for arXiv paper 2606.20757: Evidential Fusion Network for Multimodal Survival Prediction under Missing Modalities ELADO: Elliptic PDE Assessment Datasets for Operator Learning 6h ago Abstract page for arXiv paper 2606.20771: ELADO: Elliptic PDE Assessment Datasets for Operator Learning B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet 6h ago Abstract page for arXiv paper 2606.20812: B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet CELEUS: Certifiable and Efficient LLM Evaluation via E-Processes 6h ago Abstract page for arXiv paper 2606.20820: CELEUS: Certifiable and Efficient LLM Evaluation via E-Processes Evolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learning 6h ago Abstract page for arXiv paper 2606.20858: Evolutionary Discovery of Developmental Reward Schedules in Deep Reinforcement Learning Machine Learning Classification of Cryopathy Syndromes: A Comprehensive Comparative Study 6h ago Abstract page for arXiv paper 2606.20874: Machine Learning Classification of Cryopathy Syndromes: A Comprehensive Comparative Study
20 loaded
SM
stat.ML updates on arXiv.org
6h ago · 20 items
Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components 6h ago Abstract page for arXiv paper 2606.20678: Beyond Importance: Interchange-Sobol Sensitivity Reveals Task-Specific Content Channels in Transformer Components Betting on Moments: Legendre Jumper Martingales for Online Exchangeability Testing 6h ago Abstract page for arXiv paper 2606.20859: Betting on Moments: Legendre Jumper Martingales for Online Exchangeability Testing Adversarial observations in probabilistic State-Space Models for robust Reinforcement Learning 6h ago Abstract page for arXiv paper 2606.20880: Adversarial observations in probabilistic State-Space Models for robust Reinforcement Learning Diffusion-Driven State Space Models 6h ago Abstract page for arXiv paper 2606.21036: Diffusion-Driven State Space Models Bayesian Model Averaging under Predictor Redundancy via Density-Ratio Posterior Compression 6h ago Abstract page for arXiv paper 2606.21080: Bayesian Model Averaging under Predictor Redundancy via Density-Ratio Posterior Compression Two Layers of Instability in Causal Estimation 6h ago Abstract page for arXiv paper 2606.21185: Two Layers of Instability in Causal Estimation Orthogonal Discrepancy Kernels for Learning with Partial Physics 6h ago Abstract page for arXiv paper 2606.21199: Orthogonal Discrepancy Kernels for Learning with Partial Physics Subsampling for supervised learning in reproducing kernel Hilbert spaces 6h ago Abstract page for arXiv paper 2606.21260: Subsampling for supervised learning in reproducing kernel Hilbert spaces Finite-Sample Performance of Gradient Descent in Logistic Regression with Gaussian Design 6h ago Abstract page for arXiv paper 2606.21683: Finite-Sample Performance of Gradient Descent in Logistic Regression with Gaussian Design Signed Evidence Flow: Conflict-Aware and Stability-Calibrated Data Analysis 6h ago Abstract page for arXiv paper 2606.21875: Signed Evidence Flow: Conflict-Aware and Stability-Calibrated Data Analysis
20 loaded
QM
Quanta Magazine
19h ago · 5 items
A Dark Dimension Could Link Two of the Universe’s Great Unknowns 19h ago Recent observations suggest that dark energy is changing over time. Theorists wonder if dark matter is, too. Why the Human Genome’s Tangled Physicality May Confound AI 4d ago Our genetic heritage is not a blueprint or an algorithm, as many biologists have imagined, but something else entirely. Seven Perfect Shuffles Randomize a Deck of Cards. But How Many Sloppy Ones? 5d ago A decades-old proof showed that seven shuffles are enough to mix up a deck of cards. But it requires you to cut the deck with the precision of a professional magician. A new proof gets around that obstacle. How Many Elementary Particles Are There, Really? 7d ago Plausible answers range from 17 to — in all seriousness — 995.5. Where Did Earth Get Its Oceans? Maybe It Made Them Itself. 10d ago At first, scientists thought Earth’s water came from comets. Then, asteroids. Now, they wonder if Earth’s water is homegrown.
ML
Will AI spark a scientific renaissance — or a diffuse monoculture? 1d ago Artificial intelligence’s ability to enrich science will depend not only on model capability, but also on whether researchers, reviewers and funders reward originality over speed. Ovo, an open-source ecosystem for de novo protein design 3d ago The protein design field is rapidly advancing, with frequent emergence of new models and pipelines for designing de novo proteins with tailored properties and functions not found in nature. However, the current tool landscape is fragmented,... Memorization in large language models in medicine prevalence characteristics and implications 4d ago Large Language Models (LLMs) have demonstrated significant potential in medicine, with many studies adapting them through continued pretraining or fine-tuning on medical data. However, a key question remains: to what extent do LLMs memorize... A universal deep learning framework for empowering nanopore identification by reinforcing temporal signals 5d ago Nanopore sensing holds transformative potential for revolutionizing protein and glycan sequencing. However, translating this potential into practical, high-fidelity identification is severely bottlenecked by the challenge of processing mass... SplitSeek-Pro: accurate prediction of splittable sites on protein structures 5d ago Understanding protein architecture and predicting its structural tolerance to profound remodeling is pivotal for engineering functional proteins. We present SplitSeek-Pro, a deep learning model that evaluates amino acid-level splittability ... Is AI ruining our skills? Early results are in — and they’re not good 5d ago Reliance on artificial-intelligence tools degrades the abilities of physicians and software engineers, studies show. SAMJ: fast image annotation on ImageJ/Fiji via segment anything model 5d ago Accurate image annotation is essential for training artificial intelligence (AI) systems in biomedical image analysis, enabling tasks such as cell detection, tissue quantification, and disease characterization. However, creating pixel-level... Modeling visual memorability assessment with autoencoders reveals characteristics of memorable images 6d ago Image memorability refers to the phenomenon where certain images are more likely to be remembered than others. It is a quantifiable and intrinsic image attribute, defined as the likelihood of an image being remembered upon a single exposure... Reimagining machine vision with optical computing 6d ago A specialized ‘metasurface’ can preprocess incoming scene information on image-generating devices. Multi-omic analysis of deep learning-derived phenotypes links ophthalmic imaging to cardiovascular and neurological traits 7d ago The eye is a recognized source of biomarkers for cardiovascular and neurodegenerative disease risk. Here we characterize the breadth of these associations and identify biological axes that may mediate them. Using UK Biobank data, we develop...
158 loaded
FO
Future of Life Institute
3d ago · 20 items
Should AIs be people too? 3d ago Statement: Anthropic warns of AI self-improvement risks, considers a pause 15d ago FLI President on the White House Executive Order 19d ago Magnificent Humanity – The Pope’s First Encyclical Concerns AI 33d ago White House working group on AI – Statement from FLI’s Anthony Aguirre 48d ago FLI’s President and CEO on Trump’s support for an AI ‘kill switch’ 67d ago FLI CEO’s statement on the attack against Sam Altman’s home 73d ago Prominent Scientists, Faith Leaders, Policymakers and Artists Call for a Prohibition on Superintelligence, as Poll Shows Americans Don’t Want It 87d ago Statement: Head of US Policy on the White House AI legislative recommendations 92d ago Governor DeSantis Directs Florida State Agencies to Partner with Future of Life Institute to Shield Families from AI Harm 105d ago
20 loaded
AB
Ai2 Blog
5d ago · 20 items
How Domyn and AISquared built on Ai2's open releases 5d ago Domyn and AISquared show how Ai2’s open releases are helping AI labs build models for regulated industries, where transparency, provenance, licensing, and control are essential for customer trust and compliance. MolmoMotion: Language-guided 3D motion forecasting 6d ago MolmoMotion is an open, language-guided 3D motion forecasting model that predicts how object points will move in the future, enabling stronger motion prediction for robotics, video generation, and other systems that need to reason about wha... olmo-eval: An evaluation workbench for the model development loop 11d ago olmo-eval is an open evaluation workbench that helps model developers add, run, and analyze benchmarks across changing LLM checkpoints, extending OLMES from final-score reproducibility into the day-to-day model development loop. Building accessibility tools on a truly open foundation 33d ago PointCheck, an independent project, uses Molmo, MolmoWeb, and Olmo 3 to test web accessibility the way a keyboard user would—by navigating real pages and inspecting what's actually on screen. OlmoEarth v1.1: A more efficient family of models 35d ago OlmoEarth v1.1 is a more efficient family of remote-sensing models that cuts compute costs by up to 3x while maintaining similar performance, making large-scale satellite mapping faster and cheaper to run. Introducing AIMIP: The AI weather and climate model intercomparison project 41d ago AIMIP is a new open benchmark and dataset for evaluating AI climate models, showing they can match or beat conventional models on some historical climate metrics while still struggling to generalize reliably to long-term warming trends and ... Why Artificial Analysis uses Ai2's IFBench instruction-following eval 43d ago Artificial Analysis uses Ai2’s open IFBench eval because it captures a stubborn, real-world capability many benchmarks miss: whether models can reliably follow complex, multi-part user instructions. EMO: Pretraining mixture of experts for emergent modularity 46d ago EMO is a new mixture-of-experts model trained so modular expert groups emerge from data, enabling users to select small task-specific expert subsets while preserving near full-model performance. Open by design: Ai2 brings fully open AI infrastructure online with NSF OMAI 47d ago Ai2 is bringing NSF OMAI compute online to power a fully open AI research ecosystem, turning national infrastructure investment into reusable models, data, methods, and tools that can accelerate scientific discovery. MolmoAct 2: An open foundation for robots that work in the real world 49d ago MolmoAct 2 is a fully open robotics foundation model that brings faster, stronger 3D action reasoning to real-world robot tasks, alongside a major new bimanual manipulation dataset for researchers to study, reproduce, and build on.
20 loaded
GD
Google DeepMind News
6d ago · 20 items
Unlocking UK house-building with AI-accelerated planning 6d ago Google DeepMind is working alongside the UK government to co-develop an AI-powered prototype to help cut application decision times by 50%. Securing the future of AI agents 6d ago Discover our AI Control Roadmap: a defense-in-depth system to securely manage advanced, potentially misaligned AI agents. DiffusionGemma: 4x faster text generation 12d ago An overview of DiffusionGemma, an exceptionally fast text generation model with up to 4x faster speeds. Investing in multi-agent AI safety research 12d ago Google DeepMind and partners are announcing a new technical research funding call of up to $10M for researchers worldwide to strengthen multi-agent safety. Fluid, natural voice translation with Gemini 3.5 Live Translate 13d ago Gemini 3.5 Live Translate brings near real-time, natural speech translation to Google AI Studio, Google Translate and Google Meet. Introducing Gemma 4 12B: a unified, encoder-free multimodal model 13d ago An overview of Gemma 4 12B, a model designed to bring high-performance multimodal intelligence directly to your laptop. Powering the future of robotics in Europe 13d ago AI has the potential to help solve some of the world’s biggest challenges — not just in the digital realm, but in the physical world, too. Robotics is one of the most exciting frontiers of AI, where advances in language, vision and action m... Measuring the impact of learning with AI in Sierra Leone and beyond 14d ago Google DeepMind shares results from a randomized controlled trial in Sierra Leone, measuring the impact of AI in education on student learning and engagement. We’re launching the Google DeepMind Accelerator program in Asia Pacific to tackle environmental risks 32d ago The Asia-Pacific region is a global engine for economic growth, but it's also highly vulnerable to climate change. While green technologies are gaining momentum, a recen… Fast-tracking genetic leads to reverse cellular aging 35d ago Biologists Omar Abudayyeh and Jonathan Gootenberg use Co-Scientist to scan thousands of papers and identify over 20 novel factors that could reverse cellular aging.
20 loaded
TL
The latest research from Google
6d ago · 20 items
From pixels to planning: Earth AI for nature restoration 6d ago Research into how AI can help users understand skin conditions 10d ago A low-carbon computing platform from your retired phones 10d ago New framework for auditing machine unlearning 12d ago Unlocking dependable responses with Gemini Enterprise Agent Platform’s Agentic RAG 17d ago Towards passive heart health monitoring via smartphone camera 18d ago The next chapter in flood resilience: Open sourcing Google’s hydrology framework 19d ago A New Era of Discovery: Google Research at I/O 2026 25d ago Private analytics via zero-trust aggregation 26d ago Empirical Research Assistance (ERA): From Nature publication to catalyzing Computational Discovery 34d ago
20 loaded
MR
Microsoft Research
10d ago · 10 items
Ire identifies another LOTUSLITE specimen 10d ago Project Ire examined a timely malware sample and determined its intent through reverse engineering—identifying LOTUSLITE characteristics even as most major EDR tools did not detect it. Data Formulator 0.7: AI-powered data analytics for enterprise data 25d ago Data Formulator introduces AI-powered analytics for enterprise data workflows. Data teams can easily bring enterprise data into an AI-ready workspace where users can explore, analyze, and visualize data with AI agents to turn raw data into ... Extending Human Intelligence Through AI 26d ago Understanding AI as an extension of human intelligence—not a replacement for it—offers a more grounded path for building trustworthy AI systems. MagenticLite, MagenticBrain, Fara1.5: An agentic experience optimized for small models 32d ago MagenticLite is an agentic system for small models that works across the browser and local file system in a single workflow. It combines specialized models and orchestration to support efficient agentic performance on everyday tasks: Vega: Zero-knowledge proofs for digital identity in the age of AI 32d ago Vega turns a full credential into a single proof, sharing only what is needed and nothing more, with performance that works in real apps. Further Notes on Our Recent Research on AI Delegation and Long-Horizon Reliability 38d ago Our recent paper, “LLMs Corrupt Your Documents When You Delegate”, has generated discussion about the reliability of AI systems in delegated workflows. We appreciate the interest in this work and want to clarify several important points abo... mimalloc: A new, high-performance, scalable memory allocator for the modern era 40d ago mimalloc is an open-source, modern, scalable memory allocator that is a drop-in replacement for malloc and free. It is relatively small (~12K lines), with clear internal data structures, and is easy to build and integrate into other project... GridSFM: A new, small foundation model for the electric grid 40d ago Introducing GridSFM, a small foundation model that can predict AC optimal power flow in milliseconds, boosting efficiency and unlocking cost savings. Learn how GridSFM gives grid operators direct visibility into congestion, stability, and s... Advancing AI for materials with MatterSim: experimental synthesis, faster simulation, and multi-task models 41d ago MatterSim is expanding what AI can do for materials science—from faster large-scale simulations to MatterSim-MT, a new multi-task model for simulating properties beyond potential energy surfaces alone. SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests 42d ago Using SocialReasoning Bench, we observed a stable pattern across models—agents execute competently, but fail to consistently improve the user’s position, even with explicit instructions to optimize for user interest.
AS
Amazon Science homepage
12d ago · 20 items
EC2’s formally verified “isolation engine” provides mathematical assurance of virtual-machine isolation 12d ago 330,000 lines of machine-checked proofs in Isabelle/HOL verify that the Nitro Isolation Engine correctly enforces confidentiality, integrity, and memory safety between EC2 virtual machines on Graviton5. Graviton5’s improved design increases speed and energy efficiency — beyond Moore’s law 12d ago Graviton5's four-chiplet architecture, custom die-to-die connectivity, three-nanometer process, and 192 megabytes of L3 cache deliver up to 35% faster performance for web applications and ML inference. Real-world grounding in agentic AI 14d ago Physics-guided learning, calibrated uncertainty, numerical precision, and formal verification help AI agents avoid hallucinations and operate safely in warehouses, factories, and other physical systems. Bridging intent and execution in agentic systems 14d ago The harnesses that mediate between models and tools in agentic systems are becoming their own performance bottleneck, but a few simple design principles can fix what ails them. Ground truth is a process, not a dataset 19d ago A new audit-then-score protocol improves benchmark accuracy from 60.8% to 90.9% for evaluating AI fact-checkers on deep-research reports. How flat is replacing fat in AWS data center networks 25d ago A new network architecture called RNG uses passive optical ShuffleBoxes and quasi-random wiring to cut routers by 69%, boost throughput by up to 33%, and reduce network energy consumption by 40% — and it's now the default for most new AWS d... Amazon Research Awards recipients announced 26d ago Awardees represent more than 49 universities in 11 countries. Recipients have access to Amazon public datasets, along with AWS AI/ML services and tools. Diverse reasoning traces teach LLMs to make better decisions 27d ago Amazon researchers introduce set-supervised fine-tuning (SSFT) and global forking policy optimization (GFPO) to train language models that generate diverse reasoning paths — boosting single-shot accuracy on AIME 2025 and LiveCodeBench bench... Making LLMs faster without sacrificing accuracy 38d ago By treating hidden size, MLP-to-attention ratio, and grouped-query attention as first-class variables in a Chinchilla-style scaling law, researchers identify model architectures that match LLaMA-3.2 accuracy while significantly improving se... Promptimus: Improving already good LLM prompts with zero manual engineering 39d ago By focusing on specific failure points and suggesting targeted solutions, a new automated prompt-engineering framework improves prompt performance without compromising existing functionality.
20 loaded
AM
Apple Machine Learning Research
15d ago · 10 items
Introducing the Third Generation of Apple’s Foundation Models 15d ago Our next generation of Apple Intelligence is centered around our users, integrated deeply into our operating systems, and powered by a bold… IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026 26d ago Apple is presenting new research at the annual IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), which takes place in… VSAS-Bench: Real-Time Evaluation of Visual Streaming Assistant Models 32d ago Streaming vision-language models (VLMs) continuously generate responses given an instruction prompt and an online stream of input frames… EpiCache: Episodic KV Cache Management for Long-Term Conversation on Resource-Constrained Environments 35d ago Modern large language models (LLMs) extend context lengths to millions of tokens, enabling coherent, personalized responses grounded in long… BalCapRL: A Balanced Framework for RL-Based MLLM Image Captioning 43d ago Image captioning is one of the most fundamental tasks in computer vision. Owing to its open-ended nature, it has received significant… Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures 46d ago We propose HeadsUp, a scalable feed-forward method for reconstructing high-quality 3D Gaussian heads from large-scale multi-camera setups… RVPO: Risk-Sensitive Alignment via Variance Regularization 46d ago Current critic-less RLHF methods aggregate multi-objective rewards via an arithmetic mean, leaving them vulnerable to constraint neglect:… Apple Workshop on Privacy-Preserving Machine Learning & AI 2026 46d ago At Apple, we believe privacy is a fundamental human right. As AI capabilities increase and become more integrated into people’s daily… Velox: Learning Representations of 4D Geometry and Appearance 46d ago We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and… What Matters in Practical Learned Image Compression 47d ago One of the major differentiators unlocked by learned codecs relative to their hard-coded traditional counterparts is their ability to be…
NR
NVIDIA Research Unlocks Advanced Grasping, Smarter Autonomous Driving and Agent Training at Scale 19d ago New NVIDIA Research breakthroughs show how training at scale — across gripper types, driving scenarios and virtual worlds — creates AI that generalizes to diverse applications. NVIDIA Enables the Next Era Of Physical AI Research With Agent Skills For Autonomous Vehicles, Robotics And Vision AI 19d ago New physical AI agent skills, powered by NVIDIA Cosmos 3, help researchers accelerate data generation, simulation, policy training and evaluation for autonomous system development. NVIDIA Research Advances Robotics From Simulation to the Real World 25d ago Featured at the International Conference on Robotics and Automation, eight new NVIDIA Research papers show how robots trained in simulation are moving into the real world. NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather 147d ago NVIDIA Earth-2 makes weather AI accessible worldwide at every stage — from processing initial observation data to generating 15-day global forecasts or local storm forecasts. At NeurIPS, NVIDIA Advances Open Model Development for Digital and Physical AI 203d ago NVIDIA releases new AI tools for speech, safety and autonomous driving — including NVIDIA DRIVE Alpamayo-R1, the world’s first open industry-scale reasoning vision language action model for mobility — and a new independent benchmark recogni... How Do You Teach an AI Model to Reason? With Humans 299d ago NVIDIA’s data factory team creates the foundation for AI models like Cosmos Reason, which today topped the physical reasoning leaderboard on Hugging Face. NVIDIA Research Shapes Physical AI 315d ago AI and graphics research breakthroughs in neural rendering, 3D generation and world simulation power robotics, autonomous vehicles and content creation. NVIDIA Research Showcases the Future of Robotics at RSS 367d ago At this year’s Robotics: Science and Systems conference, NVIDIA Research is presenting work that advances robot learning across simulation, real-world transfer and decision-making. NVIDIA Scores Consecutive Win for End-to-End Autonomous Driving Grand Challenge at CVPR 376d ago NVIDIA was today named an Autonomous Grand Challenge winner at the Computer Vision and Pattern Recognition (CVPR) conference, held this week in Nashville, Tennessee. The announcement was made at the Embodied Intelligence for Autonomous Syst... NVIDIA Research Casts New Light on Scenes With AI-Powered Rendering for Physical AI Development 376d ago DiffusionRenderer introduces a neural rendering technique that can be applied to content generation and editing for creative fields — and to synthetic data generation for autonomous vehicles and robotics development.
18 loaded
TB
Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling 46d ago The BAIR Blog Gradient-based Planning for World Models at Longer Horizons 64d ago The BAIR Blog Identifying Interactions at Scale for LLMs 102d ago The BAIR Blog Information-Driven Design of Imaging Systems 164d ago The BAIR Blog RL without TD learning 234d ago The BAIR Blog What exactly does word2vec learn? 295d ago The BAIR Blog Whole-Body Conditioned Egocentric Video Prediction 357d ago The BAIR Blog Defending against Prompt Injection with Structured Queries (StruQ) and Preference Optimization (SecAlign) 438d ago The BAIR Blog Repurposing Protein Folding Models for Generation with Latent Diffusion 440d ago The BAIR Blog Scaling Up Reinforcement Learning for Traffic Smoothing: A 100-AV Highway Deployment 455d ago The BAIR Blog
IN
inFERENCe
117d ago · 15 items
The Future of Software 117d ago The world of software is undergoing a shift not seen since the advent of compilers in the 1970s. Compilers were the original vibe coding: they automatically generate complex machine code that human programmers had to manually write before. ... Deep Learning is Powerful Because It Makes Hard Things Easy - Reflections 10 Years On 142d ago Ten years ago this week, I wrote a post called "Deep Learning is Easy - Learn Something Harder". The post blew up, top spot on HackerNews. Needless to say, it didn't age well. Discrete Diffusion: Continuous-Time Markov Chains 397d ago A tutorial explaining some intuitions behind continuous time Markov chains for machine learners interested in discrete diffusion models. We may finally crack Maths. But should we? 1110d ago Automating mathematical theorem proving has been a long standing goal of artificial intelligence and indeed computer science. It's one of the areas I became very interested in recently. This is because I feel we may have the ingredients nee... Mortal Komputation: On Hinton's argument for superhuman AI. 1119d ago Last week in Cambridge was Hinton bonanza. He visited the university town where he was once an undergraduate in experimental psychology, and gave a series of back-to-back talks, Q&A sessions, interviews, dinners, etc. He was stopped on the ... Autoregressive Models, OOD Prompts and the Interpolation Regime 1180d ago A few years ago I was very much into maximum likelihood-based generative modeling and autoregressive models (see this, this or this). More recently, my focus shifted to characterising inductive biases of gradient-based optimization focussin... We May be Surprised Again: Why I take LLMs seriously. 1188d ago "Deep Learning is Easy, Learn something Harder" - I proclaimed in one of my early and provocative blog posts from 2016. While some observations were fair, that post is now evidence that I clearly underestimated the impact simple techniques ... Implicit Bayesian Inference in Large Language Models 1572d ago This intriguing paper kept me thinking long enough for me to I decide it's time to resurrect my blogging (I started writing this during ICLR review period, and realised it might be a good idea to wait until that's concluded) * Sang Michael ... Eastern European Guide to Writing Reference Letters 1575d ago Excruciating. One phrase I often use to describe what it's like to read reference letters for Eastern European applicants to PhD and Master's programs in Cambridge. Even objectively outstanding students often receive dull, short, factual, a... Causal inference 4: Causal Diagrams, Markov Factorization, Structural Equation Models 1838d ago This post is written with my PhD student and now guest author Patrik Reizinger [https://twitter.com/rpatrik96] and is part 4 of a series of posts on causal inference: * Part 1: Intro to causal inference and do-calculus [https://www.inferenc...
15 loaded
TG
The Gradient
124d ago · 15 items
After Orthogonality: Virtue-Ethical Agency and AI Alignment 124d ago AGI Is Not Multimodal 383d ago Shape, Symmetries, and Structure: The Changing Role of Mathematics in Machine Learning Research 583d ago What's Missing From LLM Chatbots: A Sense of Purpose 651d ago We Need Positive Visions for AI Grounded in Wellbeing 688d ago Financial Market Applications of LLMs 793d ago A Brief Overview of Gender Bias in AI 805d ago Mamba Explained 817d ago Car-GPT: Could LLMs finally make self-driving cars happen? 836d ago Do text embeddings perfectly encode text? 839d ago
15 loaded
VI
VITALab
168d ago · 10 items
Towards Brain MRI Foundation Models for the Clinic: Findings from the FOMO25 Challenge 168d ago 1. Motivation Brain Latent Progression Individual-based spatiotemporal disease progression on 3D Brain MRIs via latent diffusion 299d ago This article aims at reviewing a Alzheimer’s spatiotemporal disease progression predictive model called Brain Latent Progression (BrLP). All in all, this is ... A Survey of popular LLM Evaluation Metrics 307d ago Large Language Models (LLMs) are increasingly applied to critical domains such as medical report generation, where accuracy and trust are essential. Evaluati... Open-Source Large Language Models in Radiology: A Review and Tutorial for Practical Research and Clinical Deployment 316d ago Open-Source Large Language Models in Radiology MemSAM: Taming Segment Anything Model for Echocardiography Video Segmentation 385d ago MemSAM Simplifying Deep Temporal Difference Learning 442d ago tl;dr The authors propose PQN, a simplified deep online Q-Learning that uses very small replay buffers. Normalization and parallelized sampling from vectoriz... EchoPrime: Multi-Video View-Informed Vision-Language Model for Comprehensive Echocardiography Interpretation 456d ago Objective EchoPrime is a foundation model designed for comprehensive echocardiographic interpretation. Unlike previous models that use single views or static... DeepSeek-V3 Technical Report 497d ago DeepSeek-V3 Variational Autoencoders for Generating Synthetic Tractography-Based Bundle Templates in a Low-Data Setting 525d ago Highlights Implicit neural representations 553d ago Implicit neural networks
JM
JMLR
173d ago · 20 items
Transformers Can Overcome the Curse of Dimensionality: A Theoretical Study from an Approximation Perspective 173d ago Online Bernstein-von Mises theorem 173d ago Covariate-dependent Hierarchical Dirichlet Processes 173d ago DCatalyst: A Unified Accelerated Framework for Decentralized Optimization 173d ago Boosted Control Functions: Distribution Generalization and Invariance in Confounded Models 173d ago Contrasting Local and Global Modeling with Machine Learning and Satellite Data: A Case Study Estimating Tree Canopy Height in African Savannas 173d ago A Symplectic Analysis of Alternating Mirror Descent 173d ago Two-way Node Popularity Model for Directed and Bipartite Networks 173d ago Convergence and complexity of block majorization-minimization for constrained block-Riemannian optimization 173d ago Bayesian Inference of Contextual Bandit Policies via Empirical Likelihood 173d ago
20 loaded

No matching sources found.