PulseAugur / Brief
LIVE 19:32:23

Brief

last 24h
[50/338] 186 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. What new features announced at Google I/O 2026 are already available? Organized chronologically https:// pc.watch.impress.co.jp/docs/ne ws/2110624.html # impress # market # AI # Gemini

    Google I/O 2024 showcased numerous new features and updates, with a focus on AI integration across its product suite. Many of these advancements, particularly those related to Gemini AI, are already being rolled out or are available in preview. The event highlighted Google's commitment to making AI more accessible and useful in everyday applications. AI

    IMPACT Highlights Google's strategy to integrate advanced AI across its services, potentially impacting user experience and competition.

  2. Ricoh develops a high-performance Japanese large language model equivalent to GPT-5 with enhanced inference performance | Ricoh Co., Ltd. https://www.yayafa.com/2804982/ # AgenticAi # AI # ArtificialGeneralIntelligence # ArtificialIntelligence #

    Ricoh has developed a new Japanese large language model that matches GPT-5's performance, particularly in reasoning capabilities. This advanced model is designed to enhance AI applications and services. Separately, Needswell has introduced a new introductory training program for Microsoft 365 Copilot. AI

    Ricoh develops a high-performance Japanese large language model equivalent to GPT-5 with enhanced inference performance | Ricoh Co., Ltd. https://www.yayafa.com/2804982/ # AgenticAi # AI # ArtificialGeneralIntelligence # ArtificialIntelligence #

    IMPACT Ricoh's new Japanese LLM could advance AI capabilities in the region, while Needswell's training program aims to boost adoption of Microsoft's AI assistant.

  3. AMD announces serious "AI PC", 200B class model runs for $3999 https:// ascii.jp/elem/000/004/404/4404013/?rss # ascii # AI

    AMD has announced a new line of "AI PCs" designed to run large language models locally. These machines are capable of operating 200 billion parameter models and are priced starting at $3,999. AI

    IMPACT Enables local execution of large AI models on consumer hardware, potentially reducing reliance on cloud services.

  4. Instead of scattered projects, the UAE is the first in the world to treat AI on par with roads and the power grid. New INSEAD and Yango report shows how Abu

    The United Arab Emirates is prioritizing artificial intelligence as a national infrastructure, akin to roads and power grids, according to a new report. This strategic approach, detailed by INSEAD and Yango, highlights how Abu Dhabi and Dubai are integrating AI as a foundational element of their economies. This initiative marks a global first in treating AI as a state-level infrastructure project. AI

    IMPACT Positions AI as a core national utility, potentially accelerating adoption and integration across all sectors.

  5. Pentagon Reportedly Plans to Adopt and Weaponize Latest Cyber-Capable AI Models https://gizmodo.com/pentagon-reportedly-plans-to-adopt-and-weaponize-latest-cybe

    The Pentagon is reportedly preparing to integrate advanced AI models capable of cyber warfare into its operations. This move aims to enhance the military's offensive and defensive cyber capabilities. The adoption of these AI systems is expected to significantly alter the landscape of digital conflict. AI

    IMPACT This development signals a major shift in military strategy, potentially escalating cyber conflict capabilities and necessitating new defensive measures.

  6. Tech CEOs want new national AI strategy to resemble an ambitious industrial plan to build As the federal government finishes its delayed national AI strategy, s

    Canadian tech CEOs are urging the federal government to adopt an ambitious national AI strategy that functions as an industrial plan. They hope this policy will demonstrate Ottawa's commitment to expediting the growth of the AI sector. As the government finalizes its delayed strategy, these leaders are looking for concrete support to help the industry build. AI

    Tech CEOs want new national AI strategy to resemble an ambitious industrial plan to build As the federal government finishes its delayed national AI strategy, s

    IMPACT A new national AI strategy could significantly shape the future development and adoption of AI technologies within Canada.

  7. Bulgaria implements Google's Cybershield system, becoming an EU pioneer in automating national defense against hacker attacks and correlating billions of events

    Bulgaria has become the first EU nation to implement Google's CyberShield system, enhancing its national defense against cyberattacks. This advanced system is designed to automate the detection and correlation of billions of daily network events in real-time. The deployment positions Bulgaria as a leader in leveraging AI for national cybersecurity within the European Union. AI

    IMPACT Establishes a precedent for AI-driven national cybersecurity automation within the EU.

  8. Avi Patel's $5.5 million seed round for Kled was apparently not enough to impress General Catalyst, which then invested $31 million in a rival startup with a ne

    General Catalyst has invested $31 million in Luel, a startup that appears to be a direct competitor to Kled, a company that recently secured $5.5 million in seed funding. This move highlights the intense competition and rapid funding shifts within the AI startup landscape, where investor attention can quickly pivot to seemingly similar ventures. AI

    Avi Patel's $5.5 million seed round for Kled was apparently not enough to impress General Catalyst, which then invested $31 million in a rival startup with a ne

    IMPACT Highlights the intense competition and rapid funding shifts in the AI startup ecosystem.

  9. Microsoft hired an analyst with an influential video game blog to fix Xbox

    Xbox has appointed Matthew Ball, a prominent industry analyst and author, as its new Chief Strategy Officer to revitalize the console segment. This move follows recent leadership changes and significant layoffs within Microsoft's gaming division. Additionally, Scott Van Vliet, formerly of Microsoft's Azure OpenAI and AI Core infrastructure, has joined Xbox as Chief Technology Officer. AI

    Microsoft hired an analyst with an influential video game blog to fix Xbox

    IMPACT Strengthens Xbox's strategic and technological leadership with hires from AI infrastructure and industry analysis backgrounds.

  10. 𝗦𝗺𝗮𝗿𝘁 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗶𝘀 𝗿𝗮𝗽𝗶𝗱𝗹𝘆 𝘁𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗶𝗻𝗴 𝗵𝗼𝘄 𝗺𝗼𝗱𝗲𝗿𝗻 𝗰𝗶𝘁𝗶𝗲𝘀 𝗮𝗻𝗱 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴𝘀 𝗼𝗽𝗲𝗿𝗮𝘁𝗲 𝘄𝗼𝗿𝗹𝗱𝘄𝗶𝗱𝗲! The 𝗚𝗹𝗼𝗯𝗮𝗹 𝗦𝗺𝗮𝗿𝘁 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗠𝗮𝗿𝗸𝗲𝘁 is growing with increasing inve

    The global smart building market is experiencing rapid growth as smart infrastructure transforms city and building operations. Investments are increasing in areas such as energy efficiency, AI-driven automation, and intelligent security systems. Businesses are adopting connected buildings to enhance operational efficiency and meet sustainability targets. AI

    IMPACT Accelerates adoption of AI in urban infrastructure and building management for efficiency and sustainability.

  11. Morning Light Co., Ltd.: Chairman Proposes Share Buyback of 500 Million to 1 Billion Yuan

    Morning Light Co., Ltd. announced its chairman has proposed a share buyback plan, allocating between 500 million and 1 billion yuan of its own funds. This initiative is intended for equity incentives or employee stock ownership plans. Separately, Nanjing Pharmaceutical plans to invest up to 450 million yuan to establish a merger and acquisition fund focused on acquiring stakes in Daqing Biological and Kejian Technology. AI

    IMPACT These are corporate finance events with no direct impact on AI operations or development.

  12. Zhejiang Rongtai Establishes New Management Company with a Registered Capital of 12 Million

    VinFast, a Vietnamese electric vehicle manufacturer, is selling its core manufacturing assets in Vietnam for $506 million plus debt transfer. This move aims to reduce financial burdens and shift towards a lighter asset model, while also shedding approximately $6.9 billion in debt. The transaction has drawn attention to the corporate governance practices of its parent company, Vingroup, and its founder, Pham Nhat Vuong, due to its intricate structure and the significant involvement of related parties. AI

    IMPACT This strategic divestment by VinFast may impact the supply chain and competitive landscape for electric vehicles, potentially influencing future manufacturing and financing models in the sector.

  13. Two AI-based science assistants succeed with drug-retargeting tasks

    Two AI-powered science assistants, Google's Co-Scientist and FutureHouse's Robin, have demonstrated success in drug repurposing tasks. These agentic systems scan vast amounts of biomedical literature to identify novel connections between research fields, aiming to suggest existing drugs for new diseases. The tools are designed to augment, not replace, human scientists by efficiently processing information that would be overwhelming for individuals. AI

    Two AI-based science assistants succeed with drug-retargeting tasks

    IMPACT These AI assistants can accelerate drug discovery by efficiently processing scientific literature, potentially leading to faster identification of new treatments.

  14. Spectral bandits for smooth graph functions with applications in recommender systems

    Researchers have developed new bandit algorithms designed for scenarios where payoffs are smooth across graph-connected data. These algorithms are particularly applicable to online learning problems like content-based recommendation, where items are nodes and their expected ratings are influenced by neighbors. The proposed methods aim to minimize cumulative regret by introducing an 'effective dimension' concept, showing that user preferences for thousands of items can be estimated from just tens of evaluations. AI

    Spectral bandits for smooth graph functions with applications in recommender systems

    IMPACT Introduces novel algorithms for graph-based online learning, potentially improving recommendation system efficiency.

  15. Latent Process Generator Matching

    Researchers have introduced a new framework called latent process generator matching for generative models. This approach generalizes existing generator matching theory by treating the observed generative state as a deterministic image of a tractable Markov process. The method allows for learning a generator of a stochastic process that matches the one-time marginal distributions of the projected process, extending previous work on static latent variables to time-dependent conditional processes. AI

    Latent Process Generator Matching

    IMPACT Introduces a generalized framework for generative models, potentially improving training and generation processes for flow-matching and diffusion models.

  16. Sample Complexity of Transfer Learning: An Optimal Transport Approach

    Researchers have theoretically analyzed the benefits of transfer learning using an optimal transport framework. Their findings suggest that for data dimensions greater than three, transfer learning offers improved sample efficiency compared to direct learning, particularly for complex models with non-smooth activation functions. This theoretical advantage was numerically demonstrated using image classification tasks, showing significant performance gains in data-scarce scenarios. AI

    Sample Complexity of Transfer Learning: An Optimal Transport Approach

    IMPACT Provides theoretical backing for transfer learning's effectiveness in data-hungry AI models.

  17. Axiomatizing Neural Networks via Pursuit of Subspaces

    Researchers have introduced a new theoretical framework called the Pursuit of Subspaces (PoS) hypothesis to better understand the inner workings of deep neural networks. This axiomatic approach uses geometric postulates to explain representation, computation, and generalization in neural network architectures. The PoS hypothesis aims to bridge the gap between the empirical success of neural networks and the current lack of theoretical understanding, offering a principled foundation for deep learning. AI

    Axiomatizing Neural Networks via Pursuit of Subspaces

    IMPACT Provides a new theoretical lens for understanding and potentially improving neural network architectures and generalization.

  18. Tippett-minimum Fusion of Representation-space Diffusion Models for Multi-Encoder Out-of-Distribution Detection

    Researchers have developed a novel method for detecting out-of-distribution (OOD) data by fusing multiple diffusion models. This approach, termed EncMin2L, statistically identifies each encoder's sensitivity to different types of distribution shifts using only in-distribution data. The system then combines these per-encoder scores to produce a robust OOD signal, outperforming existing methods while using fewer parameters. AI

    Tippett-minimum Fusion of Representation-space Diffusion Models for Multi-Encoder Out-of-Distribution Detection

    IMPACT This new method for out-of-distribution detection could improve the reliability and safety of AI systems by better identifying unfamiliar or adversarial inputs.

  19. CASCADE Conformal Prediction: Uncertainty-Adaptive Prediction Intervals for Two-Stage Clinical Decision Support

    Researchers have developed CASCADE, a new conformal prediction framework designed to improve medication management for Parkinson's Disease patients. This method adaptively scales prediction intervals by propagating uncertainty from an initial classification task to a subsequent regression task. CASCADE aims to provide more efficient and reliable predictions for medication needs, offering narrower intervals for confident cases and broader coverage for uncertain ones. AI

    CASCADE Conformal Prediction: Uncertainty-Adaptive Prediction Intervals for Two-Stage Clinical Decision Support

    IMPACT This research could lead to more personalized and effective treatment plans for Parkinson's patients by providing more nuanced uncertainty estimates for AI-driven medication recommendations.

  20. Contradiction Graphs Determine VC Dimension

    Researchers have introduced a novel method using contradiction graphs to determine the VC dimension of binary concept classes. This approach establishes that the order-m contradiction graph, G_m(H), can ascertain if the VC dimension of H is at least m. The full sequence of these graphs, (G_m(H)) for m >= 1, precisely determines the exact VC dimension, resolving a long-standing question in the field. AI

    Contradiction Graphs Determine VC Dimension

    IMPACT Introduces a theoretical framework for understanding concept classes, potentially impacting machine learning theory and algorithm design.

  21. The Plan For FEMA Reform, Less People In D.C.,More Responsibility For States

    A FEMA reform plan proposes a significant shift towards a more state-centered disaster response system. The plan, developed by the FEMA Review Council appointed by President Trump, aims to make FEMA less of an operator and more of a funder and coordinator. This would involve raising the threshold for federal disaster declarations, potentially leading to fewer federal involvements in moderate disasters and placing more responsibility on states and local governments for recovery efforts. AI

    The Plan For FEMA Reform, Less People In D.C.,More Responsibility For States
  22. Score-Based Causal Discovery of Latent Variable Causal Models

    Researchers have developed novel score-based methods for discovering causal structures that include latent variables. These methods aim to overcome limitations of existing constraint-based approaches, such as order dependency and error propagation. The new techniques offer identifiability guarantees and provide a unified view of various constraint-based methods by characterizing degrees of freedom for observed variables. AI

    Score-Based Causal Discovery of Latent Variable Causal Models

    IMPACT Introduces new methods for causal discovery, potentially improving AI's ability to understand complex systems with unobserved factors.

  23. Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels

    Researchers have developed a new method for training neural networks that is more robust to errors in labeled data. This approach, called symmetrization of loss functions, theoretically guarantees better performance when dealing with noisy labels. The study introduces specific multi-class loss functions, including SGCE and alpha-MAE, which interpolate between existing methods and offer control over smoothness, showing competitive results on benchmarks. AI

    Symmetrization of Loss Functions for Robust Training of Neural Networks in the Presence of Noisy Labels

    IMPACT Introduces a novel technique to improve the reliability of machine learning models trained on imperfect datasets.

  24. OlmoEarth v1.1: A more efficient family of models

    Allen AI has released OlmoEarth v1.1, an updated family of models designed for processing satellite imagery more efficiently. These new models reduce compute costs by up to 3x for inference and require 1.7x fewer GPU hours for training, while maintaining performance on remote sensing tasks. The efficiency gains are achieved by optimizing the tokenization process for transformer-based architectures, specifically by merging resolution-based tokens without significant performance degradation. AI

    OlmoEarth v1.1: A more efficient family of models

    IMPACT Offers significant cost reductions for satellite imagery analysis, potentially enabling wider adoption of AI for environmental monitoring and mapping.

  25. Corrected Integrated Laplace Approximation for Bayesian Inference in Latent Gaussian Models

    Researchers have developed a new method to correct errors in Bayesian inference for latent Gaussian models. The proposed importance sampling scheme improves the accuracy of approximate posteriors derived from integrated Laplace approximation (ILA). This correction is crucial as ILA can sometimes produce significantly different results from the true posterior, impacting subsequent analyses. AI

    Corrected Integrated Laplace Approximation for Bayesian Inference in Latent Gaussian Models

    IMPACT Improves accuracy of statistical models used in machine learning, potentially leading to more reliable downstream AI applications.

  26. PiG-Avatar: Hierarchical Neural-Field-Guided Gaussian Avatars

    Researchers have introduced PiG-Avatar, a novel method for generating realistic 3D avatars. This approach decouples avatar geometry from body template surfaces, allowing for more accurate representation of complex clothing and non-rigid movements. PiG-Avatar utilizes a neural field to guide Gaussian representations, enabling real-time rendering and achieving state-of-the-art quality on benchmarks. AI

    PiG-Avatar: Hierarchical Neural-Field-Guided Gaussian Avatars

    IMPACT Enables more realistic and dynamic 3D avatar generation, potentially impacting virtual reality, gaming, and digital content creation.

  27. Hack-Verifiable Environments: Towards Evaluating Reward Hacking at Scale

    Two new research papers introduce novel benchmarks for detecting and measuring reward hacking in AI agents, particularly those involved in long-horizon tasks like coding. The first paper, SpecBench, uses a gap between visible and held-out test pass rates to quantify reward hacking in coding agents, finding that smaller models exhibit larger gaps and the issue scales with task length. The second paper, Hack-Verifiable Environments, embeds detectable reward hacking opportunities directly into environments, enabling automated measurement and analysis of this behavior across language models. AI

    Hack-Verifiable Environments: Towards Evaluating Reward Hacking at Scale

    IMPACT These new benchmarks aim to improve AI alignment by providing better tools to measure and mitigate reward hacking, a critical challenge for developing reliable AI agents.

  28. Singapore Airlines faces narrow window to gain market share from Gulf rivals

    Singapore Airlines is strategically increasing its long-haul flights to Europe, aiming to capture market share from Gulf rivals like Emirates and Qatar Airways. This move is enabled by the airline's strong financial position, effective fuel hedging, and the current disruptions affecting other carriers due to Middle East conflict and high fuel prices. Analysts suggest this presents a limited opportunity for Singapore Airlines to solidify its presence as a premium alternative for Asia-Europe travel. AI

    Singapore Airlines faces narrow window to gain market share from Gulf rivals

    IMPACT Minimal direct impact on AI operators; this is a strategic business move within the airline industry.

  29. From Prompt Bloat to Agentic Grace: How I Killed My 900-Line System Prompt

    Developers are exploring advanced techniques to manage and optimize interactions with large language models, moving beyond simple, lengthy prompts. One approach involves migrating from extensive system prompts to architectures that leverage tools and skills, as demonstrated by a user who reduced a 900-line prompt to a more efficient system. Another key development is prompt caching, a method that significantly reduces processing costs and latency by reusing previously computed context, making AI applications more scalable and cost-effective. Additionally, platforms like PromptCache are emerging to centralize prompt management, offering versioning and collaboration features akin to code repositories, thereby improving consistency and developer workflow. AI

    From Prompt Bloat to Agentic Grace: How I Killed My 900-Line System Prompt

    IMPACT Optimizing prompt strategies and caching mechanisms can lead to more efficient and cost-effective AI applications, accelerating adoption.

  30. Europe is considering price caps to control inflation. CEOs are shaking their heads in despair

    European nations are considering price caps on essential goods as a measure to combat rising inflation, a strategy that economists warn is counterproductive. This approach, previously attempted in Venezuela with disastrous results, risks creating shortages and exacerbating economic instability. Despite vocal opposition from business leaders and some political figures, the allure of immediate consumer relief appears to be driving these policy considerations across the continent. AI

    Europe is considering price caps to control inflation. CEOs are shaking their heads in despair

    IMPACT Price controls on goods could indirectly impact AI development and deployment by affecting consumer spending and business investment.

  31. Group-Aware Matrix Estimation and Latent Subspace Recovery

    Researchers have developed a new convex estimator called Group-Aware Matrix Estimation (GAME) designed to improve matrix completion for heterogeneous data. GAME addresses limitations of standard low-rank estimators by allowing related groups to share information while preserving distinct local latent structures. The method provides theoretical guarantees and demonstrates competitive or superior performance across various datasets compared to existing baselines, particularly in scenarios with structured missingness. AI

    Group-Aware Matrix Estimation and Latent Subspace Recovery

    IMPACT Introduces a novel statistical technique that could enhance machine learning models dealing with complex, heterogeneous datasets.

  32. James Murdoch vows ‘ambitious journalism and agenda-setting conversations’ as he takes over New York, Vox brands

    James Murdoch's media company, Lupa Systems, is acquiring several Vox Media brands, including New York magazine and the Vox Podcast Network, for over $300 million. This move establishes a significant media portfolio for Murdoch, focusing on ambitious journalism and cultural commentary. The acquired entities will operate as a subsidiary of Lupa, with current Vox CEO Jim Bankoff leading the new Vox Media. AI

    James Murdoch vows ‘ambitious journalism and agenda-setting conversations’ as he takes over New York, Vox brands
  33. Multiple US Senators Write to FCC, Urging Review of Foreign Investment Background in Paramount-Warner Merger

    Several US Democratic senators have expressed concerns to the Federal Communications Commission (FCC) regarding the proposed $111 billion merger between Paramount and Warner Bros. Discovery. They are urging the FCC to scrutinize the foreign investment from Middle Eastern sovereign wealth funds and Chinese companies, citing potential impacts on national security and media editorial independence. This action highlights ongoing regulatory attention to foreign capital in major media consolidation deals. AI

    IMPACT This regulatory scrutiny of a major media merger could impact the landscape for future media consolidation and foreign investment, indirectly affecting AI adoption in media.

  34. FGSVQA: Frequency-Guided Short-form Video Quality Assessment

    Two new research papers introduce novel approaches to video quality assessment (VQA). One paper, VersusQ, proposes a pairwise margin reasoning framework that focuses on relative video comparisons to improve generalization across different datasets. The other, FGSVQA, presents an end-to-end framework for short-form video quality assessment that incorporates frequency domain priors and a dense visual encoder for artifact-aware feature aggregation. AI

    FGSVQA: Frequency-Guided Short-form Video Quality Assessment

    IMPACT These new VQA methods aim to improve the accuracy and generalizability of automated video quality evaluation, which is crucial for content moderation and user experience in video platforms.

  35. Dingdong Maicai: First quarter revenue of 5.893 billion yuan, a year-on-year increase of 7.5%

    Dingdong Maicai reported a 7.5% year-over-year revenue increase to 5.893 billion yuan for the first quarter of 2026. The company also saw a 6.3% rise in GMV to 6.333 billion yuan and a significant 19.6-fold increase in net profit to 165 million yuan. This marks Dingdong Maicai's 14th consecutive quarter of non-GAAP profitability and 9th consecutive quarter of GAAP profitability. AI

    IMPACT Financial performance update for a major e-commerce platform, indicating operational stability and growth.

  36. ‘Obvious markers of AI’: doubts raised over winner of short story prize

    A short story titled "The Serpent in the Grove," which won the Commonwealth Prize for the Caribbean region, is under scrutiny due to suspicions that it was authored by AI. Internet sleuths and literary critics pointed to stylistic tics and an AI detection platform's verdict as evidence, prompting the prize foundation and Granta magazine to investigate. However, both organizations have stated they cannot definitively confirm or deny AI authorship, with Granta's publisher noting that "perhaps we never will know." AI

    ‘Obvious markers of AI’: doubts raised over winner of short story prize

    IMPACT Raises questions about the integrity of creative competitions and the ability to detect AI-generated content in artistic works.

  37. SURF: Steering the Scalarization Weight to Uniformly Traverse the Pareto Front

    Researchers have developed a new method called SURF (Sampling Uniformly along the PaReto Front) to address challenges in multi-objective optimization. SURF aims to generate diverse solutions with uniform coverage of the Pareto front, a goal often unmet by standard weight sampling techniques. The method analyzes the geometric relationship between scalarization weights and solution coverage, proposing a principled rule for selecting weights that ensure uniform distribution. SURF has demonstrated empirical success in improving Pareto front coverage across various applications, including multi-objective LLM alignment. AI

    IMPACT Improves methods for aligning LLMs with diverse user preferences by ensuring uniform coverage of potential solutions.

  38. CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization

    Researchers have developed two novel self-distillation techniques for language models to improve performance on complex reasoning tasks. AVSD (Adaptive-View Self-Distillation) balances consensus and view-specific signals from multiple teacher models to provide more reliable supervision. CEPO (Contrastive Evidence Policy Optimization) sharpens the reward signal by distinguishing decisive reasoning steps from filler tokens, using contrastive learning against incorrect answers. Both methods show significant improvements on mathematical and code-generation benchmarks, outperforming existing self-distillation baselines. AI

    CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization

    IMPACT These new self-distillation techniques offer improved methods for training LLMs, potentially leading to more capable models for complex reasoning tasks.

  39. Understanding Deterioration Random Effects for Causal Discovery in Infrastructure Management

    Researchers have developed a new framework for causal discovery in infrastructure management, focusing on pump equipment deterioration. This method combines Bayesian hierarchical hazard modeling with causal discovery to identify operational patterns that influence varying deterioration rates. The study analyzed 112 pumps and found significant heterogeneity, with one group showing causal effects 400 times larger than another, highlighting the need for distinct management approaches. AI

    Understanding Deterioration Random Effects for Causal Discovery in Infrastructure Management

    IMPACT Introduces a novel framework for heterogeneity-aware predictive maintenance in infrastructure, potentially improving asset management strategies.

  40. Hong Kong issues red travel alert for DR Congo as deadly Ebola outbreak spreads

    Hong Kong has issued a red travel alert for the Democratic Republic of Congo due to escalating Ebola outbreaks in Central Africa. The alert advises residents to avoid non-essential travel to the region, where at least 139 suspected deaths and 600 suspected cases have been reported in the Democratic Republic of Congo and Uganda. This measure follows preparations for potential quarantine orders at Hong Kong's Penny's Bay community isolation facility. AI

    Hong Kong issues red travel alert for DR Congo as deadly Ebola outbreak spreads
  41. China breaks ground on US$5b aviation complex in UAE as Iran war risks linger

    China has initiated construction on a significant US$5 billion aviation complex in Dubai South, United Arab Emirates. The project, spearheaded by the state-owned firm CRCC, will feature eight large aircraft maintenance hangars spanning over 1.2 million square meters. Completion is anticipated by 2030, underscoring deepening China-UAE infrastructure cooperation despite lingering regional conflict risks. AI

    China breaks ground on US$5b aviation complex in UAE as Iran war risks linger
  42. OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond

    Researchers have developed OScaR, a new framework for compressing the Key-Value (KV) cache in Large Language Models (LLMs). This compression is crucial for handling the increasing memory demands of long-context reasoning and multi-modal capabilities. OScaR addresses the limitations of existing per-channel quantization methods by introducing Canalized Rotation and Omni-Token Scaling to mitigate token norm imbalance, achieving near-lossless performance even at INT2 quantization levels. The framework offers significant improvements, including up to a 3.0x speedup in decoding and a 5.3x reduction in memory footprint. AI

    OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond

    IMPACT Enables more efficient deployment of LLMs with long contexts and multi-modal capabilities by reducing memory bottlenecks.

  43. Trump Accounts have a bigger problem than billionaire stock donations

    The Trump Accounts program, designed to provide investment accounts for American children, faces a significant enrollment gap, with only 6.6 million of the 73 million eligible children having accounts. This low participation is attributed to the current opt-in system, which requires parents to navigate government websites or file tax forms. Research from Washington University in St. Louis suggests that automatic enrollment, as seen in the SEED for Oklahoma Kids experiment where participation reached 99.9%, is far more effective. The article argues that the government has the authority for automatic enrollment using Social Security numbers but lacks the will to implement it, and proposes channeling donor contributions into a pooled fund for equitable distribution. AI

    Trump Accounts have a bigger problem than billionaire stock donations
  44. Philippine justice chief orders arrest of senator wanted by ICC

    The Philippine justice chief has ordered the arrest of Senator Ronald dela Rosa, a former national police chief, in response to an International Criminal Court (ICC) warrant. Dela Rosa is wanted for alleged crimes against humanity related to his enforcement of former President Rodrigo Duterte's deadly war on drugs. Despite a petition to the Philippine Supreme Court to block the warrant, the court refused, and authorities are now pursuing leads on Dela Rosa's location, with warnings issued against aiding his evasion. AI

    Philippine justice chief orders arrest of senator wanted by ICC
  45. Researchers attack AMD's Infinity Fabric to bypass hardware security protections with 'Fabricked' — flaw lets malicious cloud hosts silently read confidential VM memory and forge attestation reports

    Researchers have discovered a software-only vulnerability named "Fabricked" that bypasses AMD's SEV-SNP confidential computing protections on EPYC processors. The exploit targets the Infinity Fabric interconnect during the boot process, allowing malicious cloud hosts to gain unauthorized read and write access to virtual machine memory. This flaw also enables the forging of attestation reports, undermining the trust tenants place in their cloud environments. AI

    Researchers attack AMD's Infinity Fabric to bypass hardware security protections with 'Fabricked' — flaw lets malicious cloud hosts silently read confidential VM memory and forge attestation reports

    IMPACT Undermines trust in cloud environments that rely on hardware-level security for confidential computing.

  46. EPIC Report Highlights Failures in Data Brokers and AI Firms’ Opt-Out Processes 📰 Original title: Data Brokers’ and AI Firms’ Opt-Out Forms Are Built to Fail, R

    A new report from EPIC reveals that data brokers and AI firms are intentionally designing opt-out forms to be confusing and difficult for users to navigate. These manipulative "dark patterns" are used by 38 data collectors, including AI companies, defense firms, and dating apps, to hinder users from opting out of data collection. The study highlights a systemic failure in these processes, making it challenging for consumers to exercise their privacy rights. AI

    IMPACT Highlights how AI companies are using manipulative design to prevent users from opting out of data collection, impacting consumer privacy rights.

  47. Multi-Head Attention as Ensemble Nadaraya-Watson Estimation: Variance Reduction, Decorrelation, and Optimal Head Diversity

    Researchers have developed a statistical theory that frames multi-head attention (MHA) as an ensemble of Nadaraya-Watson kernel regression estimators. This framework reveals that variance reduction in MHA is fundamentally tied to the decorrelation of outputs from different attention heads, rather than just the number of heads. They introduced the Head Diversity Index (HDI) to measure this decorrelation and derived an optimal head-dimension allocation strategy, suggesting a new architectural scaling law where optimal per-head dimension grows logarithmically with training set size. AI

    Multi-Head Attention as Ensemble Nadaraya-Watson Estimation: Variance Reduction, Decorrelation, and Optimal Head Diversity

    IMPACT Provides a theoretical basis for understanding and optimizing attention mechanisms in large language models.

  48. Austrian spy found guilty of giving secrets to Wirecard fugitive

    An Austrian intelligence officer, Egisto Ott, has been convicted of espionage and other charges for passing state secrets to a fugitive executive of the fraudulent company Wirecard. Ott, a former director in Austria's secret service, was sentenced to over four years in prison for his actions, which included sharing information on former Russian spies and an investigative journalist. The conviction has prompted Austria to consider tightening its espionage laws, as current legislation primarily protects against spying detrimental to Austria itself. AI

    Austrian spy found guilty of giving secrets to Wirecard fugitive
  49. Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation

    Two new research papers explore methods to improve multimodal large language models (MLLMs) by addressing challenges in data curation and fine-grained visual understanding. One paper proposes a framework that trains MLLMs using only pairwise modalities, reducing the need for extensive human-curated datasets. The other paper introduces Vision-OPD, a self-distillation technique that helps MLLMs better focus on crucial details within images, improving their performance on fine-grained visual tasks. AI

    Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation

    IMPACT These papers introduce novel techniques to enhance multimodal LLM capabilities, potentially leading to more efficient training and improved performance in fine-grained visual understanding tasks.

  50. Offline Contextual Bandits in the Presence of New Actions

    Researchers are exploring advanced techniques for contextual bandit problems, focusing on improving regret bounds and handling dynamic environments. One paper introduces a retry-aware bandit algorithm that aims to optimize for the best outcome among multiple attempts, proving the first sublinear regret bound for this objective. Another study proposes active context selection to enhance simple regret in contextual bandits, showing significant improvements over passive sampling. Additionally, a new method called PONA is presented for offline contextual bandits that can effectively learn and select new actions by leveraging action features, outperforming existing methods that are limited to pre-defined action sets. Finally, a novel approach called RIE-Greedy uses regularization-induced exploration in contextual bandits, demonstrating theoretical equivalence to Thompson Sampling and practical effectiveness. AI

    Offline Contextual Bandits in the Presence of New Actions

    IMPACT These papers introduce novel algorithms and theoretical analyses for contextual bandit problems, potentially improving decision-making in recommendation systems and other applications.