Results for “models”

arXiv (research papers) Yuxi Zheng

Beyond the Sink Interpretation: Radial Attention in Flow-Based VLA Policies

Prior work calls the high attention on the first and last action tokens of a flow-based VLA policy a radial action sink, and uses these tokens to set the execution horizon. In language models, a sink is a token with high incoming attention, a small value, and little effect on the output. We test whether that profile holds, not the horizon rule, in one $π_{0.5}$ checkpoint on five RoboCasa tasks. The two endpoints receive 5.4% of action-key mass, 1.4 percentage points above…

Read at arXiv (research papers)

arXiv (research papers) Ziyue Wang

Alignment Drives Reuse: Learning Generalizable Robot-Use Agents from Heterogeneous Experience

Vision-language models (VLMs) demonstrate broad generalization across tasks and domains. However, comparable generalization in robot manipulation remains challenging. Heterogeneity in robot embodiments, task formulations, geometric conventions, and action interfaces fragments interaction experience across benchmarks and hardware. We rethink robot skill learning through alignment: Can heterogeneous interactions across benchmarks and embodiments jointly improve one shared…

Read at arXiv (research papers)

arXiv (research papers) Saharnaz Mehrani Security

When Suspicion Adds Up: A Stochastic Framework for Optimizing Honeypot-Based Insider-Threat Detection

Insider threats pose security risks when employees or other trusted users misuse their access to organizational systems. Detecting these threats is difficult because harmful activity can resemble routine behavior. Honeypots, or decoy resources monitored for suspicious interactions, can provide evidence of potentially malicious activity. We develop a decision framework that models evidence accumulation through honeypot interactions using a continuous-time Markov chain and…

Read at arXiv (research papers)

arXiv (research papers) Zilang Cen

Reasoning on Call: Efficient Agentic Robot Control for Long-Horizon Robot Manipulation

Large multimodal models can support long-horizon robot manipulation, but frequent model calls incur substantial latency and token costs. The central challenge is determining when agent assistance or intervention is needed and whether errors require correcting robot execution, subtask planning, or both. To address this problem, we propose RoC (Reasoning on Call), an efficient framework that coordinates a low-level policy and a high-level reasoning agent through an online…

Read at arXiv (research papers)

Android Authority Aamir Siddiqui Gadgets

Samsung prepares Galaxy AI on Snapdragon X Elite for its upcoming Android laptops

One UI 9.5 Beta 1 reveals local photo editing, Gauss language models, and 8K video support for Samsung's Origin laptops.

Read at Android Authority

arXiv (research papers) Xuankun Rong Security

DailySafe: Benchmarking Dynamic Safety Awareness in Everyday Multi-Turn Multimodal Interactions

As multimodal large language models (MLLMs) become increasingly integrated into everyday life, their ability to provide safe and reliable assistance in realistic daily activities remains insufficiently explored. Existing safety benchmarks primarily focus on isolated harmful requests or adversarial scenarios, overlooking the safety risks that may arise as ordinary tasks unfold. In practice, users pursue legitimate goals through multi-turn, multimodal interactions, where…

Read at arXiv (research papers)

arXiv (research papers) Changtao Zhao

FastRoverVLA: Fast Vision-Language-Action Execution for Off-Road Autonomous Driving

Off-road vision-language-action (VLA) models generate scene-conditioned plans, but frequent trajectory updates can repeatedly invoke expensive multimodal inference. Reusing a prediction reduces this demand while leaving its commands exposed to changes in vehicle motion. FastRoverVLA addresses this mismatch by decoupling visual plan generation from motion-conditioned execution. Its training-free interface maintains decoded plans between visual predictions. A speed-curvature…

Read at arXiv (research papers)

arXiv (research papers) Zheyuan Liu

Teaching Vision-Language-Action Models to Stop Safely

Vision-language-action (VLA) models enable robots to perform diverse tasks from visual observations and language instructions. While prior work has explored task suppression and refusal, stopping prohibited behavior once execution is underway and controlling the resulting physical response remain challenging. Our evaluation reveals that policies which suppress a task from its initial state can still complete it when taking over partway through execution. Even when the task…

Read at arXiv (research papers)

arXiv (research papers) Wei Dai

EmbodyAgent: Post-Training Embodied VLM Agents via Distillation, SFT, and RL

Vision-language models (VLMs) can act as closed-loop robot-use agents (RUAs) by writing and executing control code through tool interfaces. However, turning an open-weight VLM into a high-performing embodied agent requires a complete post-training stack: out-of-the-box open VLMs fail completely at interactive tool contracts, strong frontier teachers are accessible only as black-box endpoints without exposed weights or token probabilities, and supervised imitation alone…

Read at arXiv (research papers)

arXiv (research papers) Jinxi Liu

One View, One Action: Spatially Generalizable VLAs through Observation-Aligned Experts and Cross Graph Consistency

Multi-view vision-language-action (VLA) models still face challenges in spatial generalization, particularly under changes in camera viewpoints and robot initial states. These models typically receive action supervision only in the robot base frame, requiring them to implicitly learn the spatial correspondence between observations from each view and base-frame actions. We propose One View, One Action (OVOA-VLA), which learns to predict actions in each view's camera frame…

Read at arXiv (research papers)

Android Authority Ankit Banerjee Gadgets

Aurzen Black Friday projector deals: Up to 40% off Roku TV and Google TV models

Whether for family movie night or to catch the game with friends, Aurzen has you covered.

Read at Android Authority

arXiv (research papers) Jiawei Xing Security

ChronoPatch: Black-Box Energy-Latency Attacks on Vision-Language Models

Vision-language models (VLMs) are increasingly deployed as generative inference services, where autoregressive decoding makes response length a key determinant of latency and computational cost. This creates an availability risk: adversarial visual inputs can prolong generation and increase serving overhead. Existing energy-latency attacks typically rely on gradients, logits, or token-level feedback, leaving open whether such risks persist when an attacker observes only the…

Read at arXiv (research papers)

VietnamPlus (Vietnamese state media) Geopolitics

President of Azerbaijan and spouse begin official visit to Vietnam

President of Azerbaijan Ilham Aliyev’s visit carries particular significance. It offers high-ranking leaders a chance to reinforce political trust, review the full range of relationship-building directions and set concrete steps to realise the Strategic Partnership, adding new momentum across politics, diplomacy, economy, trade, investment, energy, education, science-technology, and people-to-people exchanges. Vietnam introduces community-based tourism models at Azerbaijan…

Read at VietnamPlus (Vietnamese state media)

The Decoder Maximilian Schreiner SI

OpenAI's own researchers now say they value their AI tools more than nearly half their colleagues

OpenAI's internal use of coding agents is growing fast, with theoretical token costs now exceeding researchers' base salaries. Even though teams are reluctant to work without these tools, the models still produce a lot of "AI slop" beyond routine tasks. The article OpenAI's own researchers now say they value their AI tools more than nearly half their colleagues appeared first on The Decoder.

Read at The Decoder

SiliconANGLE AI Jonathan Anthony SI

Enterprises fear AI success more than failure, because success means lock-in they can’t escape

The supercomputing world built its reputation on scientific simulation, but the artificial intelligence infrastructure now filling the same data centers raises a harder question. When models turn data into revenue, control over that data and the intelligence it produces becomes a board-level concern few companies discuss publicly. A new buyer’s guide sets out to measure

Read at SiliconANGLE AI

arXiv (research papers) John Won

Learning What Actions Change: Action Effect Matching for Dynamics-Faithful World Models

World models aim to predict environment dynamics under actions, enabling the prediction of action consequences without executing them in the real world. Yet despite producing increasingly realistic videos, current world models often fail to capture how small action changes affect future observations. This limits their use in tasks such as robotic policy evaluation and model-based planning, where nearly identical actions can lead to qualitatively different outcomes. To…

Read at arXiv (research papers)

arXiv (research papers) Robin Geens

Kameleo: In-Situ Activation-Precision Selection for Bit-Serial LLM Accelerators

Weight-only quantization keeps the activations of large language models (LLMs) in a higher-precision format, pushing the on-chip inference cost to the activation arithmetic. Bit-serial, variable-length accelerators offer a promising path to reduce this cost by lowering activation precision while preserving model quality. They process only the mantissa bits that each layer needs. The length has to be chosen per layer and operator, and today it is chosen by offline…

Read at arXiv (research papers)

The Quantum Insider Matt Swayne

BMO Reports Quantum Utility in Machine Learning Application Using IBM Quantum Computers

Insider Brief PRESS RELEASE — BMO today announced it has demonstrated quantum utility in a machine learning application using IBM quantum computers, meaning the work was performed reliably at a scale beyond what exact traditional computing methods can practically reproduce. The work explores how quantum computing could help AI models identify which information matters most

Read at The Quantum Insider

arXiv (research papers) Francisco W. Kerche Tech policy

Web search in Latin America: language models use less local and diverse news than search engines

AI platforms are becoming a major gateway to news and current events. This shift risks amplifying existing geographical biases that favour narratives from wealthier countries over those from the Global South. This is particularly relevant in Latin America, which accounts for the largest share of AI usage in the Global South. We examine how information about Latin America changes when users shift from conventional search engines to LLM-based search. We construct a dataset of…

Read at arXiv (research papers)

arXiv (research papers) Luca Colombo

Matching Architecture to Workload: Comparing RISC-V Parallel Accelerators for Energy-Efficient Embedded Computing

Embedded computing demands high performance and energy efficiency across workloads with widely different computational characteristics. Choosing the most suitable execution model therefore requires a quantitative understanding of how each model maps to different workload classes. In this paper, we compare three execution models for energy-efficient embedded computing: scalar multicore, stream ing multicore, and vector processing. We use three open-source RISC-V accelerators…

Read at arXiv (research papers)

Android Authority Aamir Siddiqui Gadgets

Exclusive: Samsung’s upcoming 14-inch and 16-inch Android laptops are codenamed ‘Origin’

One UI 9.5 beta code reveals Wi-Fi-only XE940 and XE960 models powered by Qualcomm's Snapdragon X Elite chip.

Read at Android Authority

arXiv (research papers) Yao Zhang

Fast-JEPA: Advancing the Success-Latency Frontier for Latent World Models

Latent world models predict the outcomes of action sequences, but comparing candidate plans through recursive rollout remains expensive at test time. We introduce Fast-JEPA, a latent world model planning framework that advances the success-latency Pareto frontier of predictive search. Fast-JEPA reorganizes predictive computation to support effective action-sequence search at low planning cost. It constructs a goal-directed plan to guide search and directly predicts the…

Read at arXiv (research papers)

arXiv (research papers) Youssef ElNahas Tech policy

Verifiable Logic in Medical LLMs: A Critical Review of Agentic Clinical Reasoning Optimization and Neuro-Symbolic Grounding

Clinical large language models are usually judged by whether the final answer is right. Yet a model can reach a fluent, plausible output through invalid reasoning that static benchmarks do not detect, and confident hallucination, causally invalid inference, and spurious demographic shortcuts can pass conventional evaluation. This critical review covers four paradigms in turn: how clinical hallucination arises and why static metrics miss it; the agentic hallucination cascade…

Read at arXiv (research papers)

arXiv (research papers) Xinzhuang Xiong

Meta-RTL: Evolving Agent Harnesses for RTL Optimization

Large language models enable agents to optimize RTL using EDA feedback, yet their executable search procedures remain largely hand- engineered. We present Meta-RTL, an EDA-grounded framework that evolves RTL optimization harnesses from a minimal loop while keeping the optimization model, EDA flow, and per-design budgets fixed. Compared with Dr. RTL, Meta-RTL* reduces aggregate WNS/TNS violations by 7.8%/11.0% across 20 development designs, with area reductions on 70% of…

Read at arXiv (research papers)

ReliefWeb HALO Trust

Quarter of a Million Landmines Removed from Zimbabwe - The HALO Trust has destroyed four mines per person in north-east Zimbabwe

Countries: Zimbabwe, Afghanistan, Angola, Colombia, Sri Lanka, Syrian Arab Republic, Ukraine, Yemen Source: HALO Trust Harare, Zimbabwe, 19 Nov. The world’s largest landmine clearance charity, the HALO Trust, has reached a significant milestone in Zimbabwe, destroying 250,000 landmines since 2013. Dense belts of landmines, including the highly sensitive R2M2 models, were laid along Zimbabwe’s borders with Mozambique and Zambia during the Liberation War in the 1970s. They’ve…

Read at ReliefWeb

Help Net Security Industry News Security

Palo Alto Networks to acquire Applied Compute for $4.5 billion

Palo Alto Networks will acquire Applied Compute, a platform for post-training and inference of open-weight AI models. The all-stock transaction is valued at $4.5 billion, subject to adjustments. The acquisition will bring Applied Compute’s platform and world-class AI research team to Palo Alto Networks to help enterprises adopt AI securely, whether they choose open-weight models, closed frontier models, or both. Palo Alto Networks believes the future of enterprise AI is…

Read at Help Net Security

Tech.eu John Reynolds Startups

London-based Boltz raises $110M Series A

London-based AI drug discovery startup Boltz has raised a $110m Series A round, less than a year after emerging out of stealth.Boltz is an AI research startup that develops AI models to speed up drug ...

Read at Tech.eu

Analytics India Magazine Ridhika Basnet SI

Can Simulated Users Replace Weeks of Testing Products? Vectorial AI Thinks So

The startup is targeting businesses in India, the US, and Canada with AI-generated audience simulations. It plans to expand its behavioural models and product

Read at Analytics India Magazine

Digital Trends Vikhyaat Vivek

Microsoft’s new AI coder is gunning for Claude and ChatGPT. It’s coming with a price advantage

Microsoft has launched two homegrown AI coding models that promise performance comparable to expensive OpenAI and Anthropic alternatives at a fraction of the cost, with both already available in GitHub Copilot.

Read at Digital Trends

arXiv (research papers) Pironti Francesco Aurelio Security

ALIoTh: Orchestrating Multi-Agent LLMs for Autonomous IoT Penetration Testing

Penetration testing in Internet of Things (IoT) environments, especially those composed of various different devices, e.g., smart plugs, smart cameras, and smart doors, presents great challenges across the phases of information gathering, attack surface enumeration, privilege escalation, and lateral movement. This paper introduces ALIoTh, a multi-agent, workflow-driven penetration testing framework that integrates Large Language Models agents within a…

Read at arXiv (research papers)

Older stories