Artificial Intelligence
Artificial Intelligence news — Monday 26 October 2026
150 stories from 11 sources, every one in English. Each links to its publisher.
-
Anthropic maintains its focus on safety in IPO filing
AI Business
Anthropic structured its corporate voting power in a special LLC to maintain its mission of building safe AI technology.
-
AI learning experiment without breakdown, 4000 times faster in heat,...1.76 hours→1.6 seconds
AI Times (Korea)
Translated from Korean
In public experiments learning language model without breakdown, time to reach target performance decreased from 1.76 hours to 1.6 seconds. The open source community improves learning algorithms and GPU integration. Although it’s not…
-
RTX 30 Graphic Card ‘Unlocked’ tool...AI Image and video processing up to 60% increase
AI Times (Korea)
Translated from Korean
An open source tool that pushes some AI operational restrictions on the GeForce 'RTX 30' series graphics card in NVIDIA. In user testing, image/video generation processing increased by about 60%. If you do not buy a new graphics card, you…
-
"Exhibition" with 1024 AI collaboration without MS and administrators...
AI Times (Korea)
Translated from Korean
Microsoft (MS) has released the system to open source collaboration with 1024 AI coding agents without central administrators. In a way that agents choose their work and share their results, the size of the cursor code review and the role…
-
Honda demonstrates remote-controlled robot for complex physical tasks
AI Business
The system illustrates how remote human control could help robotics developers deploy machines before AI can reliably handle complex physical tasks on its own.
-
Hegging Face, DNA Reads and Answers to ‘Carbon-X’ Reveals...Electronic Analysis Cost Low
AI Times (Korea)
Translated from Korean
When hegging face enters DNA sequence, he revealed the AI model ‘Carbon-X’, which predicts the impact of the gene change and explains the reason. In self-evaluating, large-scale general-purpose AI has greatly reduced the cost of churon. It…
-
Meta, ‘Muse’ feature magnification speed... Ready to share Avatars in real time voice conversations
AI Times (Korea)
Translated from Korean
Meta will provide real-time voice conversations to the AI agent ‘Muse’ for personalization, and we also have an animation Avatar sharing feature. We use cookies to ensure that we give you the best experience on our website. If you continue…
-
Will the real Anthropic please stand up
MIT Technology Review Biotech
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here . After weeks of frenzied speculation, Anthropic has filed its IPO prospectus. The 335-page document…
-
We are home to military silence measures... From address inquiry to feedback submission
AI Times (Korea)
Translated from Korean
-
OpenAI, discounted and detropicable customers...andtropicy is a new model to lead use
AI Times (Korea)
Translated from Korean
OpenAI’s price discount for codecs and API models, and we’re going to take a free credit forward to the expenditure customer’s spending. In some businesses, the use of two company models was reversed. After launching and launching new…
-
Logo creation Beyond SVG Editing File... Design Practices with Audiogram, Vector AI
AI Times (Korea)
Translated from Korean
AI image creation Competition is expanding to the stage of creating files that can be utilized in real design tasks in a step that makes a good picture. We use cookies to ensure that we give you the best experience on our website. If you…
-
[Board] Wontid Lab, FDE Job Fair, Makina Rocks, Besui Participation, etc.
AI Times (Korea)
Translated from Korean
■ The One-Ted Lab (CEO Lee Tae-gi) announced that it will hold the ‘FDE’ recurring carnival at Lotte World Tower, Seoul, on the 18th of the following month. This year, the number of companies that have been FDE accreditation was 13.8 times…
-
2nd place nationwide with 20.2% of total sales and growth companies
AI Times (Korea)
Translated from Korean
The second largest company in the country in Korea, the sales high-growing company. In recent three years, the company’s total sales increased by 10% per year, and the total number of companies increased by 20% per year. According to the…
-
Gwangju Bus, Renamed Routes in 9 Years... Enhanced New Residential Buildings from 31
AI Times (Korea)
Translated from Korean
From the 31st to the 31st of Gwangju city buses will be operated by the new route system. The routes to connect major destinations such as the new residential complex and public transport area, and reduce the routes to return. It is a…
-
Criminal prediction AI learning up to the frontier of the police.. . "The vicious circle of the car"
AI Times (Korea)
Translated from Korean
AI predicts the possibility of crime, while it is used to cope with police, the polarized short-speed record of the past police has been raised to AI, and it is likely to be able to discriminate against specific areas and minorities. AI is…
-
Copyright © 2017. All Rights Reserved.
AI Times (Korea)
Translated from Korean
Gwangyang-si uses fossil fuel to promote the supply of heating equipment to the air heat pump. It is a plan to draw heat during the air and use it for heating and hot water to reduce gas or oil use. According to the ‘Gwangyang-style green…
-
Copyright (c) 2015 ILSHIN TECH. All Rights Reserved.
AI Times (Korea)
Translated from Korean
In the future, we will continue to develop the AI semiconductor industry in Korea. Gyeongbukdo has signed a ‘MOU’ for the establishment of ‘Gyeongbuk AI Semiconductor System and Industrial Ecosystems, such as Gyeongsangbuk IT Convergence…
-
8.4 billion dollars export of agricultural products in South Korea...
AI Times (Korea)
Translated from Korean
Kim has been the largest number of items in the export of agricultural products in Korea. Kim, who won 41.5% of the total export amount, only the agricultural products exported to the United States, were 72.6%. According to the 25th…
-
Dumb&Partners invested in StarCloud, the U.S. Space Data Center
AI Times (Korea)
Translated from Korean
Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager, Deputy General Manager…
-
Elvis Songs 274 songs ... Antropicture, KRW 540 billion copyright litigation
AI Times (Korea)
Translated from Korean
American Pop’s legendary Elvis Fresley’s music rights raised copyright infringement litigation against AI entrepreneurial andropic. Antropic is a claim that 274 songs are used for AI model learning, and the results that infringe copyright…
-
Jinon Launches AI Guard Solution for Industry
AI Times (Korea)
Translated from Korean
We will notify you if you have any questions or concerns regarding our products or services. If you have any questions about our products, please feel free to contact us. We developed a standalone solution based on the experience of…
-
China AI Computing Enduro...65% Rely on American Chip
AI Times (Korea)
Translated from Korean
The AI computing resources that China can use are doubled up to the end of the year, but about 65% of the entire U.S. enterprises have come out to rely on the chip designed. In addition to the production of semiconductors in China, the…
-
Establishment of AI Bio-Safety Research Institute, the world's first. .17 billion investment
AI Times (Korea)
Translated from Korean
The U.S. government establishes the National AI Biosecurity Institute (AIBL) to detect and monitor biological threats using AI. Incorporation of research institutes, we are planning to cooperate with major AI companies such as OpenAI…
-
Inside the push to make workers trust and use AI
AI Business
Enterprises use educational resources, communication and special initiatives to ease employees into accepting AI tools. They also face some resistance from employees.
-
What Makes AI-Written Stories Feel Off?
Towards Data Science
And how we can turn that feeling into insights The post What Makes AI-Written Stories Feel Off? appeared first on Towards Data Science .
-
Altman “Academics, AI’s ‘Beautiful Over’ Waters in Dosss”
AI Times (Korea)
Translated from Korean
Sam Altman OpenAI CEO said that mathematics who have impacted AI’s mathematics research achievements have just a few days of receiving the skills and began finding their roles. If you’re looking for an AI, you’ll need to set up a problem…
-
Launched ‘Cellstar’, a decision-making AI.. . Increase your judgment ability and reduce costs
AI Times (Korea)
Translated from Korean
Select a page Select a page... id="menu-item-15">Home id="menu-item-1768">Past Issues id="menu-item-2447">Book Reviews id="menu-item-1081">About User Guide "Apache 2.0" license for commercial use and modification and reproduction. Dabi is…
-
Opening the Box: Can Machine Learning Resolve Schrödinger’s Cat?
Towards Data Science
Neural networks won’t settle the measurement problem — but they are quietly changing how we see, predict and protect quantum superpositions. The post Opening the Box: Can Machine Learning Resolve Schrödinger’s Cat? appeared first on…
-
“Advertising Show” Guidelines on Doorshots, Influencers... “Kimi” Shinjae Promotion Exhibition
AI Times (Korea)
Translated from Korean
The doorshot AI has been deployed as a result of the latest model ‘Kimi K3.1 Pro’ to hide advertisements from overseas influencers. The public guidance has been approved prior to posting content, and the paid partnerships are not…
-
Clarification Samsung Hospital, Park, Soo - Jin Bae Yong-ICU visitation privileges, however..
AI Times (Korea)
Translated from Korean
Andropic’s ‘Callload’ revealed that it approaches the U.S. Georgian vote-related data without human approval. As the state government confirmed to bypass the payment barriers in the process of accessing the voting aggregate information…
-
Certain Fertility Treatments Linked To Kawasaki Disease Risk In Children
Asian Scientist
A study from Japan found higher rates of Kawasaki disease among children conceived through ovulation induction, intrauterine insemination and IVF, though the findings do not establish a cause-and-effect relationship.
-
Launched ‘XGEN TMS’, an AI test management solution for in-house closed network
AI Times (Korea)
Translated from Korean
Platform (CEO) announced that it launches the AI-based testing management solution ‘XGEN TMS’ that can be utilized in a corporate closed network environment. XGEN TMS is a solution that integrates quality verification processes from…
-
Seoul AI hub, launching ‘Bubby Yunicon Center’
AI Times (Korea)
Translated from Korean
Seoul AI hub, an AI hub, supported by AI startups in Seoul, announced that it will lead to the new creation of ‘Bubby Yunicon Center’, which will help the expansion of AI startups succeeded in investing. It is a ward to create a base that…
-
Investigation, Investigation, Investigation, Investigation .'Body's Tracking
AI Times (Korea)
Translated from Korean
The Financial Supervisor has built a big data analysis system that finds the after-sales organization with an individually accepted illegal financial damage report. This system has been exposed to damages that were buried, and the…
-
NVIDIA, $2000 per GPU SW subscription... Ecosystem ‘Lock’ Acceleration
AI Times (Korea)
Translated from Korean
NVIDIA announced that the software required to drive AI agent was a recognition of $2,000 per GPU (about $264) per year. A strategy to expand your business structure to a subscription-type software to secure revenue. In order to improve…
-
The Download: everyday AI harms and the race to nuclear spaceflight
MIT Technology Review Biotech
This is today’s edition of The Download , our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Opinion: The focus on AI safety disasters ignores everyday harms —Sam Hiner, cofounder of the Young…
-
Import AI 478: Math for alignment; ProgramBench; and Mirendil shows what RSI means
Import AI Newsletter
Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now How math could help alignment: …What the…
-
My Machine Learning Pipeline Cheated, and Cross-Validation Never Noticed
Towards Data Science
What an ID column, a date string, and a missing value taught me. The post My Machine Learning Pipeline Cheated, and Cross-Validation Never Noticed appeared first on Towards Data Science .
-
Secure GPUs with space X, 53.. "Neocloud is a crissor"
AI Times (Korea)
Translated from Korean
SpaceX completed $400 billion (about $53 trillion) debt procurement to secure the GPU required for AI operations. Sunon Musk CEO emphasized the foundation of large-scale funding, the revenue generated by StarLink satellite Internet and…
-
OpenAI will host ‘Model Masters’ for Korean startups. .11 Exhibitors
AI Times (Korea)
Translated from Korean
OpenAI announced that it was held ‘openAI model masters’ to support the AI service development of domestic startups. Model Masters is a program that supports startups to implement AI utilization ideas into real products and features, and…
-
U.S. Housing Solar Price Lift...Powered by Battery Combined Equipment
AI Times (Korea)
Translated from Korean
US Home Solar Installation Quotation Price has been increased by 3.2% in the first half of the year. If only solar is installed, the price did not change, but the solar part price of the equipment to install with the battery was about 5%…
-
Registered a 6.2MWp business for the European Solar and ESS trading platform
AI Times (Korea)
Translated from Korean
The European Solar and Energy Storage System (ESS) project trading platform was expanded to the Romanian market. Solmeria, a French company, was registered as a promising two business combined with zero-tolerant solar and battery in…
-
China's solar industry, the modular price has been pushed...It is still low
AI Times (Korea)
Translated from Korean
China’s major solar companies discussed the implementation of a module selling price, but the actual export price has maintained the same level for two consecutive weeks. The price suggested in the industry is higher than the current…
-
Claude Code vs. Cursor: What a Field Test Shows That the Feature Comparisons Miss
Towards Data Science
One benchmark says Claude Code used 33,000 tokens where Cursor used 188,000. Another says the exact reverse. Here is the field test that actually settles it for your team. The post Claude Code vs. Cursor: What a Field Test Shows That the…
-
ChatGPT’s expanding world of work
OpenAI Blog
Explore OpenAI Signals data on 165 work activities across nearly 100 countries, revealing how people use ChatGPT for work and how usage changes over time.
-
The focus on AI safety disasters ignores everyday harms
MIT Technology Review Biotech
Consider what passes for success in technology policy. An AI company promises that its chatbot will not encourage suicide . A social media platform makes it more difficult for children to buy fentanyl . These are important protections, and…
-
[AIReader] AI Autonomous Learning AI Agent Development
AI Times (Korea)
Translated from Korean
AI Autonomous Learning AI Agent Development Introductory | On July 1, 2026, we recommend this book from LLM’s foundation to AI agent, and we implement core technologies directly through the practice of using the OpenAI API. The reader can…
-
How Ultrafast cut incident investigation from hours to minutes at OpenAI
OpenAI Blog
See how OpenAI engineers use Codex Ultrafast to investigate incidents in minutes. Explore faster security analysis, code refactoring, and daily workflows.
-
How Fanatics and OpenAI streamline the fan experience
OpenAI Blog
Fanatics uses OpenAI to help fans complete their game-day outfits with Shop the Look. Testing across roughly 30,000 MLB products showed 10% higher engagement.
-
PoKER: A Probability of Kill Estimation Rate Model for Air-to-Air Missiles Using Machine Learning on Stochastic Targets
arXiv cs.LG
arXiv:2610.32864v1 Announce Type: new Abstract: This work introduces PoKER, a novel probabilistic model engineered to optimize missile launch effectiveness in air-to-air scenarios, specifically within Beyond Visual Range (BVR) air combat…
-
ImageRSI: Verification-Driven Recursive Self-Improvement for Agentic Image Generation
arXiv cs.CV
arXiv:2610.32120v1 Announce Type: new Abstract: Complex text-to-image requests remain difficult to satisfy reliably. Verifying and repairing an image can fix an individual output, but a successful repair does not necessarily carry over to…
-
Searching with Sparse Histories: Cross-Task Architecture Search from Incomplete Search Logs and Zero-Cost Proxies
arXiv cs.AI
arXiv:2610.32179v1 Announce Type: cross Abstract: Neural architecture search (NAS) is expensive, but cross-task transfer can reduce costs by reusing knowledge from previous tasks. However, existing methods struggle with sparse search…
-
NeuroMAS: Multi-Agent Systems as Neural Networks with Joint Reinforcement Learning
arXiv cs.AI
arXiv:2605.16757v2 Announce Type: replace Abstract: Multi-agent language systems are often built as hand-designed workflows, where agents are assigned semantic roles and communication protocols are specified in advance. We propose…
-
Effectiveness of Curriculum Learning Depends on Reward Sparsity and Competing Optima
arXiv cs.LG
arXiv:2610.31884v1 Announce Type: new Abstract: While animals and people tend to learn a task more quickly and reliably when they first train on simpler versions of it, curriculum learning's effectiveness in artificial settings varies, and…
-
REACT: Representational Early Analysis and Control of Chain-of-Thought Trajectories
arXiv cs.CL
arXiv:2610.32113v1 Announce Type: cross Abstract: Chain-of-thought (CoT) generation offers an opportunity to intervene before a reasoning model completes its answer. But does the point at which risk is easiest to detect also offer the most…
-
Nested Sparsity for Scalable Pretraining and Continual Learning
arXiv cs.LG
arXiv:2610.31818v1 Announce Type: new Abstract: Mixture-of-experts (MoE) transformers, the standard backbone for modern large-scale deep learning, employ conditional computation: only a sparse subset of model weights is used in each…
-
Rethinking LoRA Initialization for Robust Asymmetric Learning Rates
arXiv cs.AI
arXiv:2610.32142v1 Announce Type: cross Abstract: Low-Rank Adaptation (LoRA) is often highly sensitive to learning-rate (LR) choices. Recent empirical studies suggest that, once LRs are carefully tuned, vanilla LoRA can remain competitive…
-
When and How Does Muon Precondition? The Role of Structured Gradient Noise
arXiv cs.LG
arXiv:2610.31728v1 Announce Type: new Abstract: Muon has become a popular optimizer for large language model training, yet its theoretical foundations remain poorly understood. In this work, we study how gradient noise geometry shapes…
-
Many Perturbations, Shared Failure Dynamics: Tracing Direction-Sensitive Perturbation Influence in Language Models
arXiv cs.CL
arXiv:2610.32309v1 Announce Type: new Abstract: Why can a task-preserving rewrite turn a correct answer into a failure while a larger internal change leaves the answer intact? We develop a direction-sensitive, readout-relative account of…
-
Acoustic and Lexical Signals for Native Language Identification of Language Learners
arXiv cs.CL
arXiv:2610.32382v1 Announce Type: new Abstract: Spoken native language (L1) identification for second language learners has traditionally relied on acoustic attributes of speech. Comparatively little attention has been paid to analysing…
-
KrishokTech: A Bengali Agricultural Advisory System with Explicit Authorization Controls
arXiv cs.CL
arXiv:2610.32817v1 Announce Type: new Abstract: For agricultural advisory, generating an answer is only part of the problem: the system must also control unsafe requests, work with limited connectivity, and accommodate changing guidance…
-
Not All Retained Clients Are Equal: Client Selection for Federated Unlearning
arXiv cs.LG
arXiv:2610.31786v1 Announce Type: new Abstract: Federated unlearning (FU) aims to remove a designated target client's influence from a trained model without the prohibitive cost of retraining from scratch. Existing methods largely treat…
-
Stochastic Inverse Problem: stability, regularization and Wasserstein gradient flow
arXiv cs.LG
arXiv:2410.00229v2 Announce Type: replace-cross Abstract: Inverse problems in physical or biological sciences often involve recovering an unknown parameter that is random. The sought-after quantity is a probability distribution of the…
-
Tradeoff-Aware Territory Design: A Multi-Objective Decision System for Enterprise Customer-to-Seller Allocation
arXiv cs.AI
arXiv:2610.31701v1 Announce Type: new Abstract: Enterprise customer journeys are shaped not only by product experiences and marketing, but also by recurring resource-allocation decisions. In B2B go-to-market (GTM) organizations, territory…
-
Rewriting Scaling Laws for Synthetic Data with Compute and Data Constraints
arXiv cs.CL
arXiv:2610.31867v1 Announce Type: cross Abstract: Synthetic rewrites, where a language model rewrites a natural document e.g. into a target style, are now a standard ingredient of pretraining corpora. However, critical aspects of their use…
-
ViT-H: Generative Video-To-Humanoid Control with Motion Tracking Feedback
arXiv cs.CV
arXiv:2610.32966v1 Announce Type: cross Abstract: Learning humanoid control from monocular human videos offers a scalable way to acquire diverse whole-body behaviors. Yet, existing reconstruction-retargeting-tracking pipelines are limited…
-
Residual Expert Routing for Text-Only Domain Adaptation of Factorized Transducer ASR
arXiv cs.CL
arXiv:2610.32637v1 Announce Type: new Abstract: Fine-tuning the non-blank predictor of a factorized transducer on target-domain text improves recognition on that domain but degrades performance on the source domain. A single interpolation…
-
Spectra: Multi-Scale Implicit Decomposition for Deepfake Detection
arXiv cs.AI
arXiv:2610.31979v1 Announce Type: cross Abstract: Deepfakes increasingly undermine trust in digital media, yet existing detectors often struggle to generalize across unseen manipulation techniques, compression levels, and dataset shifts…
-
Earth embeddings are decomposable records of the land surface
arXiv cs.LG
arXiv:2610.32617v1 Announce Type: new Abstract: Ecological monitoring, carbon accounting and biodiversity assessment increasingly rely on annual Earth embeddings--learned vectors that compress a year of satellite observation for every…
-
End-to-End Differentiable Diffusion Conditioning for Physics-Informed Optimization
arXiv cs.AI
arXiv:2610.31721v1 Announce Type: cross Abstract: Optimizing high-dimensional physical systems under expensive evaluations and complex constraints is a central challenge in science and engineering. In offline model-based optimization…
-
Order Matters: Competition-Guided Query Ordering for RNN-Based Object Detection
arXiv cs.CV
arXiv:2610.05191v3 Announce Type: replace Abstract: DETR-style detectors use one-to-one bipartite matching during training to assign object queries to ground-truth objects, enabling end-to-end set prediction without non-maximum suppression…
-
The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows
arXiv cs.AI
arXiv:2610.31985v1 Announce Type: new Abstract: LLM agents are increasingly deployed in collaborative settings where they pursue separate tasks while sharing resources and operating under review or oversight processes. Such settings create…
-
BiLoCo: Binary Low-Rank Corrections for Post-Training Quantization
arXiv cs.LG
arXiv:2610.32243v1 Announce Type: new Abstract: Post-training quantization (PTQ) to FP4 has emerged as a key technique for reducing large language model inference cost, especially on Blackwell GPUs, which offer native FP4 tensor core…
-
Minimum Description Length Unsupervisedly Separates Law-Grade from Statistical-Grade Knowledge in a Self-Improving Science Agent - and Reveals 86% of Its Own Discoveries Are Statistical
arXiv cs.AI
arXiv:2610.32245v1 Announce Type: new Abstract: AI systems that discover scientific relations now generate candidate "laws" at scale, but the bottleneck has shifted from generation to vetting: which candidates deserve belief? We show that…
-
STEP: Surgical Point Tracking from Sparse Endpoints and Pseudo-trajectories
arXiv cs.CV
arXiv:2610.31951v1 Announce Type: new Abstract: Surgical point tracking requires persistent correspondence through tissue deformation and occlusion, yet dense trajectory annotations are costly to obtain. We present STEP, a two-stage…
-
On the Runway Cascade of Transformers for Language Modeling
arXiv cs.LG
arXiv:2601.14522v2 Announce Type: replace Abstract: In decoder-only (causal) transformers, the computation graph created by causal masking routes information through both direct-path attention and indirect paths formed by intermediate…
-
Let's Do Turbulence Right: In Space and Time!
arXiv cs.LG
arXiv:2610.32601v1 Announce Type: cross Abstract: The three-dimensional Navier-Stokes equations govern fluid turbulence, a chaotic and non-linear physical phenomenon characterized by complex eddies interacting across vast spatial and…
-
IGG: A Benchmark for Interactive GUI Grounding under Visibility Constraints
arXiv cs.AI
arXiv:2610.32536v1 Announce Type: new Abstract: GUI grounding benchmarks ask whether an agent can localize a target that is already visible in a static screenshot. Real interfaces often violate this assumption: a target may be off-screen…
-
Romanized Arabic Across Dialects: Views, Usage Patterns, and Linguistic Variation
arXiv cs.CL
arXiv:2608.02555v2 Announce Type: replace Abstract: Arabizi refers to Arabic written in Latin script. Although previous studies have shown that the prevalence and usage of Arabizi vary by factors such as region and age group, most NLP…
-
Uni-Stream: A Unified Streaming Speech Language Model for Speech Synthesis and Recognition
arXiv cs.AI
arXiv:2610.32187v1 Announce Type: cross Abstract: Unified streaming speech language models aim to support speech recognition and synthesis within a single model, enabling real-time human-computer interaction. However, jointly supporting…
-
Dual-Adaptive Z-Score PPO: Dynamic Clipping Using Action Confidence and Normalized Advantage Magnitude
arXiv cs.LG
arXiv:2610.32194v1 Announce Type: new Abstract: Proximal Policy Optimization (PPO) remains a popular on-policy reinforcement learning algorithm owing to its stability and ease of implementation. Its clipping threshold, however, is fixed…
-
Track finding and particle identification performance with particle physics foundation models
arXiv cs.LG
arXiv:2610.31180v2 Announce Type: replace-cross Abstract: Modern high energy nuclear and particle physics experiments produce large amounts of sparse data that must be reduced significantly in size before performing physics analysis…
-
CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition
arXiv cs.AI
arXiv:2610.32022v1 Announce Type: new Abstract: Intent recognition in human-robot interaction requires agents to resolve ambiguous natural language instructions by inferring the speaker's underlying goals. Research has increasingly focused…
-
SOL: Measuring Gaps between Text Distributions by Double Sliced Wasserstein Metrics
arXiv cs.LG
arXiv:2610.06513v3 Announce Type: replace-cross Abstract: Evaluating text generation requires measuring how well the generated distribution matches the data distribution. For autoregressive models, this is done by the perplexity. Diffusion…
-
Topological Equivalence of Spectral and Density Cluster Hierarchies
arXiv cs.LG
arXiv:2610.31909v1 Announce Type: new Abstract: We compare two canonical ways to recover hierarchical clustering from an underlying density $f$: the \emph{density cluster tree} given by connected components of superlevel sets…
-
Don't Forget What You Just Learned: Self-Directed Repair Turns Agent Mistakes Into Lasting Lessons
arXiv cs.AI
arXiv:2610.31846v1 Announce Type: new Abstract: Intelligent agents should learn from their mistakes so that experience leads to continuous improvement. Yet today's agents recover from a mistake only temporarily: in-context reflections are…
-
Localization and Neural Splatting on the Edge: An FPGA Scene Memory for Robots
arXiv cs.CV
arXiv:2610.25068v2 Announce Type: replace-cross Abstract: Resource-constrained robots and autonomous vehicles need scene memories that localize a camera and render nearby views without a desktop GPU. Scene-coordinate regression (SCR) and…
-
SemantEq: Faithful Bidirectional Equivalence of Formal Statements
arXiv cs.AI
arXiv:2610.32919v1 Announce Type: new Abstract: Statement autoformalization translates mathematics from natural language into a proof assistant such as Lean 4, and models now do it at scale. As provers get stronger, the burden of…
-
Random Jump Forests
arXiv cs.LG
arXiv:2610.32498v1 Announce Type: cross Abstract: Random jump forests are introduced as a nonparametric tool for estimating conditional cumulative transition rates and occupation probabilities in multi-state models. The main idea is that a…
-
Correct but Unselectable: The Hidden Interface Tax in Multi-Candidate Reasoning
arXiv cs.LG
arXiv:2610.32235v1 Announce Type: new Abstract: Small language models deployed under cost, latency, or privacy constraints cannot simply switch to larger models when reasoning fails. Multi-candidate generate-then-rerank offers a natural…
-
Searching with Sparse Histories: Cross-Task Architecture Search from Incomplete Search Logs and Zero-Cost Proxies
arXiv cs.LG
arXiv:2610.32179v1 Announce Type: new Abstract: Neural architecture search (NAS) is expensive, but cross-task transfer can reduce costs by reusing knowledge from previous tasks. However, existing methods struggle with sparse search…
-
Measure Before You Train: Typed Decision Models vs. a Classical Baseline vs. a Generative LLM for Routing a Personal Multilingual Inbox
arXiv cs.LG
arXiv:2610.32955v1 Announce Type: cross Abstract: Personal email agents decide, for every message, what it is, whether it is urgent, whether to draft a reply, create a task, or ignore it. Their error rate is rarely measured: the stored…
-
Maximum Likelihood on the Joint (Data, Condition) Distribution for Solving Ill-Posed Problems with Conditional Flow Models
arXiv cs.LG
arXiv:2208.11782v2 Announce Type: replace Abstract: In the absence of conditioning information, normalizing flow models are trained using maximum likelihood. Previous extensions to conditional settings have abandoned the simple…
-
Reuse, Don't Retrain: Efficient, Modality-Versatile Geospatial Foundation Models through Feature Distillation
arXiv cs.CV
arXiv:2610.32566v1 Announce Type: new Abstract: Geospatial foundation models (GeoFMs) are costly to pretrain, yet model families often repeat this process across backbone scales and sensing modalities. We investigate feature distillation…
-
FlexGlue: Sparse-to-Dense Feature Matching and Visual Disambiguation
arXiv cs.CV
arXiv:2610.32976v1 Announce Type: new Abstract: To estimate correspondences across images, existing approaches match either sparse interest points, which is highly efficient but limited by repeatability, or dense pixel grids, which is…
-
Test-Time Scaling of Text-to-Image Flow Matching Models with Noise-Rejuvenated Sequential Monte Carlo
arXiv cs.CV
arXiv:2610.25479v2 Announce Type: replace Abstract: Test-time scaling of diffusion models with sequential Monte Carlo (SMC) has become a promising strategy for sampling under user-specified rewards, such as human preference scores…
-
StreamEval: Evaluating Interactive Streaming Video Generation Models
arXiv cs.CV
arXiv:2610.28402v2 Announce Type: replace Abstract: We introduce StreamEval with four key features: (1) First interactive streaming video generation benchmark. (2) Fully open source. Benchmark data, evaluation code, protocols, and scoring…
-
A survey of diversity quantification in natural language processing: The why, what, where and how
arXiv cs.CL
arXiv:2507.20858v4 Announce Type: replace Abstract: The concept of diversity received increasing attention in natural language processing (NLP) in recent years. It became an advocated property of datasets and systems, with many measures…
-
Training-Free Task Vectors for LLM Behavioral Control
arXiv cs.LG
arXiv:2609.09054v3 Announce Type: replace Abstract: Task vectors enable post-training model editing by identifying semantically meaningful directions in weight space, typically computed as the difference between a fine-tuned model and its…
-
Bandits with Probing: Optimal Coverage and the Role of Feedback
arXiv cs.LG
arXiv:2609.15248v2 Announce Type: replace Abstract: In bandits with probing, a learner selects at most $k$ of $n$ arms per round, earns the largest of their rewards, and competes with the best fixed single arm. For independent arms with…
-
CFBrainAug: Modularity-Induced Counterfactual Brain Network Augmentation for Interpretable fMRI-based Autism Diagnosis
arXiv cs.LG
arXiv:2610.32184v1 Announce Type: new Abstract: Machine learning on resting-state functional magnetic resonance imaging (rs-fMRI) brain networks is promising for diagnosing Autism Spectrum Disorder (ASD), but existing methods are prone to…
-
What Transfers from Training-Only Priors? Separating Knowledge, Ordinal Targets, and Optimization
arXiv cs.LG
arXiv:2610.11146v3 Announce Type: replace Abstract: Training-only teacher access lets a compact student deploy with local features and exported weights. This established distillation interface does not identify what transfers or when it…
-
Collaborative In-Context Learning for Indoor Radio Map Estimation
arXiv cs.LG
arXiv:2610.31702v1 Announce Type: new Abstract: Indoor radio map estimation (RME) is essential for wireless applications such as coverage planning, localization, and network deployment. However, most existing methods follow a supervised…
-
Where Rounding Happens Matters: Rounding-Location Alignment Stabilizes LLM Reinforcement Learning
arXiv cs.LG
arXiv:2610.32951v1 Announce Type: new Abstract: Reinforcement learning for large language models often uses separate systems for inference and training. Even with identical weights and BF16 tensors, implementation differences can make an…
-
Accordion: Elastic-Depth Inference via Fixed-Prefix Layer Looping
arXiv cs.CL
arXiv:2610.32492v1 Announce Type: cross Abstract: Large language models serve applications with diverse inference budgets, but training and maintaining separate models for each budget incurs substantial costs. Recent Looped Transformers…
-
FinVerse: Toward More Realistic Evaluation for Financial Time-Series Forecasting
arXiv cs.LG
arXiv:2608.03259v3 Announce Type: replace Abstract: Time-series foundation models (TSFMs) are increasingly evaluated on broad benchmarks that apply uniform conventional forecast evaluation metrics across heterogeneous time-series. However…
-
Behaviour analysis for large language model-driven agents
arXiv cs.AI
arXiv:2610.32087v1 Announce Type: new Abstract: Understanding language-agent behaviour requires learning how actions are organised across a trajectory and how this organisation relates to context and outcomes. We propose a general…
-
On Orthogonal Recurrent Neural Networks Quantization
arXiv cs.LG
arXiv:2402.04012v3 Announce Type: replace-cross Abstract: In recent years, Orthogonal Recurrent Neural Networks (ORNNs) have gained popularity due to their ability to manage tasks involving long-term dependencies, such as the Copy-task…
-
Astro2Method: Benchmarking LLMs on Machine Learning Method Selection in Astronomy
arXiv cs.AI
arXiv:2610.30282v1 Announce Type: cross Abstract: Machine learning (ML) is increasingly used in astronomy, but choosing a suitable method for a scientific problem often requires expertise in both astronomy and ML. Large language models…
-
BrainRotViT: Transformer-ResNet Hybrid for Explainable Modeling of Brain Aging from 3D sMRI
arXiv cs.LG
arXiv:2511.15188v3 Announce Type: replace-cross Abstract: Accurate brain-age estimation from structural MRI provides a valuable biomarker for studying healthy aging and neurodegenerative disease. We propose BrainRotViT (Brain ResNet over…
-
OlmoEarth v1.3: Embeddings
arXiv cs.CV
arXiv:2610.32294v1 Announce Type: new Abstract: We make some OlmoEarth embeddings. To do this, we train a new model: OlmoEarth v1.3. OlmoEarth v1.3 is designed to condense multimodal, multitemporal remote sensing data into a single…
-
Data Quality in RVL-CDIP: A Systematic Audit
arXiv cs.CV
arXiv:2610.32126v1 Announce Type: new Abstract: RVL-CDIP is the de facto standard benchmark for document classification, and reported accuracy on it is now 98.8%. In this paper, we audit RVL-CDIP for data quality issues and find that…
-
Multilevel Stochastic Plug-and-Play for Sparse-View CT Reconstruction
arXiv cs.CV
arXiv:2606.24567v3 Announce Type: replace Abstract: Sparse-view computed tomography (SVCT) reduces radiation exposure and acquisition time, but the limited number of projection views makes the reconstruction problem severely ill-posed and…
-
Concentrated Causal Influence, Distributed Diagnostic Information: Cognitive Impairment Decisions in Large Audio-Language Models
arXiv cs.CL
arXiv:2610.32061v1 Announce Type: cross Abstract: Large audio-language models (LALMs) can assess cognitive impairment (CI) from speech, yet their answers reveal neither which internal components control the decision nor how much CI…
-
Learning to Predict, Discover, and Reason in High-Cardinality Event Sequences
arXiv cs.LG
arXiv:2603.16313v4 Announce Type: replace-cross Abstract: Electronic control units (ECUs) embedded within modern vehicles generate a large number of asynchronous events known as diagnostic trouble codes (DTCs). These discrete events form…
-
Watermarking Should Be Treated as a Monitoring Primitive
arXiv cs.LG
arXiv:2605.13095v4 Announce Type: replace-cross Abstract: Watermarking is a primary means for providers of generative models to meet legal requirements to mark synthetic content. A watermark embeds a hidden signal into generated content…
-
Pseudo-Labels as Rewards, Not Targets: Group-Relative Policy Optimization for Semi-Supervised ASR
arXiv cs.CL
arXiv:2610.32572v1 Announce Type: new Abstract: Adapting ASR to a domain with only untranscribed audio typically relies on pseudo-labeling, a strong teacher transcribes the audio and a student is fine-tuned on it. Because supervised…
-
Characterize Then Distill: Mechanistic Reasoning in Large Output Spaces
arXiv cs.LG
arXiv:2606.06840v3 Announce Type: replace-cross Abstract: Reasoning-trained language models can select, zero-shot, a small set of relevant labels from thousands to hundreds of thousands of candidates. We ask how they do it mechanistically…
-
DBCM: Dual-Branch Calibration Module for Test-Time Adaptation in Time Series Forecasting
arXiv cs.LG
arXiv:2610.32071v1 Announce Type: new Abstract: Non-stationarity induces test-time distribution shifts in time series forecasting (TSF), resulting in forecast residuals between predictions and the corresponding ground truth. Recent works…
-
Depth-Dependent Safety Geometry in Vision-Language Models: Layer-Wise Localization and Bidirectional Refusal Control
arXiv cs.LG
arXiv:2610.32182v1 Announce Type: cross Abstract: Vision-Language Models (VLMs) are widely deployed, yet the internal mechanisms behind their safety behavior remain poorly understood. We present a lightweight inference-time framework for…
-
RootDown: Deploying and Benchmarking Models for Beet-Root Detection on Edge Devices
arXiv cs.LG
arXiv:2610.32275v1 Announce Type: cross Abstract: Quality inspection of sugar beets after harvest and storage is critical to preserve sugar content and ensure optimal processing efficiency. The transfer of harvested beets to field clamps…
-
Decoupling candidate dual AGN from chance superpositions in the GOTHIC survey via a deep-learning framework
arXiv cs.LG
arXiv:2608.24164v3 Announce Type: replace-cross Abstract: Dual active galactic nuclei (DAGN) mark a critical phase in the evolution of merging galaxies and the pairing of supermassive black holes, yet they remain difficult to identify in…
-
If at First You Don't Succeed, Backtrack: Lightweight Recovery for Frozen VLAs
arXiv cs.LG
arXiv:2610.32941v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models perform well on robotic manipulation, yet remain brittle once their own actions carry them into states their training data barely covers. Recovering from…
-
How Should Transformers Encode Numeric Values in Electronic Health Records?
arXiv cs.LG
arXiv:2607.01391v2 Announce Type: replace Abstract: How do we encode numeric values in transformer-based sequence processing, particularly in electronic health record (EHR) data? We systematically compare discrete, continuous, and hybrid…
-
Recurrent Self-Improvement: Dynamic Cross-Loop On-Policy Distillation for Looped Language Models
arXiv cs.LG
arXiv:2610.10623v2 Announce Type: replace Abstract: Looped Language Models (LoopLMs) offer a parameter efficient approach to scaling reasoning by reusing shared parameters across recurrent computation steps. Despite their promise…
-
Efficient Omni-modal Large Language Model Inference via Per-Query Modality Selection
arXiv cs.LG
arXiv:2610.31768v1 Announce Type: new Abstract: Omni-modal large language models (Omni-LLMs) can jointly process a broad range of modalities, such as text, audio, and video, enabling stronger multimodal understanding and reasoning…
-
Latent space configuration for neural network finetuning, fast ranking, and embedding dimensionality reduction
arXiv cs.AI
arXiv:2610.29295v2 Announce Type: replace-cross Abstract: Finding K most similar elements for a query in a large gallery of neural network (NN) embeddings is essential in various areas of computer science and industrial applications of…
-
The Price of Current-Effect Inference in Nonstationary Multi-Armed Bandits
arXiv cs.LG
arXiv:2610.29982v2 Announce Type: replace-cross Abstract: Adaptive experiments use bandit algorithms to favor interventions with higher expected rewards. However, decision-makers also require reliable inference on current treatment…
-
The LLENS Platform for On-Premises Large Language Model Agents in a Japanese University Hospital: Development and Evaluation Study
arXiv cs.CL
arXiv:2610.32151v1 Announce Type: new Abstract: Large language model agents that use tools could take on part of the electronic health record and desk work that occupies physicians' time, but in many hospitals the medical information…
-
Which Answer Did the Seventeen-Year-Old Write? A Behavioral Case Study of Author-Age Misattribution, Miscalibrated Confidence, and Non-Corrective Self-Explanation in Six Frontier Language Models
arXiv cs.CL
arXiv:2610.31946v1 Announce Type: new Abstract: Three students answered the same exam prompt by hand: "What is happiness? Are you happy?" Their ages, recorded at collection, are 17, 28, and 45. The 17-year-old's answer is hedged and…
-
Rethinking Multi-View Fusion for Urban Representation Learning
arXiv cs.AI
arXiv:2610.31850v1 Announce Type: cross Abstract: Urban region representation learning encodes heterogeneous urban signals, such as mobility flows, points of interest, and land-use patterns, into unified embeddings for downstream tasks…
-
SGF+: Decoupling Gradient Flows for Autoregressive Video Generation
arXiv cs.CV
arXiv:2610.10429v3 Announce Type: replace Abstract: Autoregressive video generation requires denoising the current frames while writing their key-value representations as context for future predictions. However, these two roles typically…
-
STROKE: Simple Sparse Structural Occlusions for Black-Box Adversarial Attacks
arXiv cs.CV
arXiv:2610.32932v1 Announce Type: new Abstract: We introduce STROKE, a query-efficient score-based black-box adversarial attack that constructs perturbations from simple oriented rectangular occlusions. STROKE optimizes these perturbations…
-
ParaTaxe: a Suite of Tools for Processing and Publishing Multi-parallel Corpora
arXiv cs.CL
arXiv:2610.32796v1 Announce Type: new Abstract: Over the past fifteen years, digital tools for parallelizing and aligning textual corpora have developed significantly within the field of digital humanities. Several projects and software…
-
Validating the Comparison, Not Just the Scorer: Why Overall Accuracy Can Mislead in Interactive LLM Evaluation
arXiv cs.AI
arXiv:2610.31678v1 Announce Type: cross Abstract: Many LLM studies compare how models respond to different follow-up prompts. For example, one prompt may neutrally ask a model to reconsider, while another insists that an invented film…
-
How Do I Know What to Say Next? Barenholtz's Autogenerative Theory as an Enrichment of Harrisean Integrationism
arXiv cs.CL
arXiv:2607.07891v2 Announce Type: replace Abstract: Roy Harris's Integrationist linguistics offers a compelling critique of the referentialist tradition embedded deep at the heart of computational approaches to language, arguing that…
-
Knowledge Distillation across Item and Code Spaces for Parallel Generative Recommendation
arXiv cs.LG
arXiv:2610.32943v1 Announce Type: cross Abstract: Generative recommendation with semantic IDs represents each item as a sequence of discrete codes derived from its semantic representation and predicts these codes to generate…
-
The Role of Quantum in Hybrid Quantum-Classical Neural Networks: An Empirical Assessment
arXiv cs.AI
arXiv:2601.04732v2 Announce Type: replace-cross Abstract: Quantum machine learning has emerged as a promising application domain for near-term quantum hardware, particularly through hybrid quantum-classical models that leverage both…
-
POSEIDON: Scaling Training Environments for Scientific Agents from Code Repositories
arXiv cs.AI
arXiv:2610.32448v1 Announce Type: new Abstract: Training environments remain a major bottleneck for agents in data-driven scientific discovery. Existing collections of executable and verifiable training environments are small, and their…
-
On the Limits of Stochastic Push-Pull over Time-Varying Digraphs
arXiv cs.LG
arXiv:2610.32422v1 Announce Type: cross Abstract: Stochastic Push-Pull achieves linear speedup with respect to the network size $n$ on static strongly connected digraphs, but whether this speedup persists on time-varying networks remains…
-
MPQ: A Message-Passing View of Post-Training Quantization
arXiv cs.AI
arXiv:2610.32127v1 Announce Type: new Abstract: Post-training quantization (PTQ) is a key technique for efficiently deploying large language models, as it compresses weights into low-bit representations without retraining. Existing PTQ…
-
The Butterfly Defects: Tracing Cooperation in LLM Agent Networks
arXiv cs.CL
arXiv:2610.32726v1 Announce Type: cross Abstract: As large language models (LLMs) increasingly operate as interconnected agents, a fundamental question emerges: does individual cooperation translate into collective cooperation? We…
-
AlphaAnything: Extending Pretrained RGB Generators to RGBA
arXiv cs.CV
arXiv:2610.32413v1 Announce Type: new Abstract: Existing RGBA generation methods commonly learn a transparency-specific representation through reconstruction, introducing an additional training stage and potentially altering the RGB…
-
SANE: Automated Scorer-Specific Metamorphic Testing for LLM Judges
arXiv cs.AI
arXiv:2610.28886v2 Announce Type: replace-cross Abstract: LLM judges increasingly mediate benchmark evaluation, model selection, data curation, and training. Their decisions are useful only when each judge responds reliably to changes…
-
Effectiveness of Curriculum Learning Depends on Reward Sparsity and Competing Optima
arXiv cs.AI
arXiv:2610.31884v1 Announce Type: cross Abstract: While animals and people tend to learn a task more quickly and reliably when they first train on simpler versions of it, curriculum learning's effectiveness in artificial settings varies…
-
Where Cross-Question Structure in LLM Errors Comes From: Shared Seeds, Shared Prompts, and Answer-Letter Preference
arXiv cs.AI
arXiv:2607.20464v2 Announce Type: replace Abstract: Disagreement between language models is often read as missing knowledge. Questions that need the same missing knowledge should then be right or wrong together. We test for this structure…
-
Approximating velocity fields with planted point attractors via Neural-ODE for classification purposes
arXiv cs.LG
arXiv:2606.23550v3 Announce Type: replace-cross Abstract: In this work, Neural ODEs equipped with a curated collection of equilibrium points have been successfully employed for classification tasks. The planted fixed points serve as…
-
Phronix: The Capability-Centric Learning Paradigm for Agents
arXiv cs.AI
arXiv:2610.31789v1 Announce Type: new Abstract: Scientific computing increasingly relies on large language model-based agents for performance engineering tasks such as instrumentation, analysis, and workflow porting. However, agent…
-
From Deliberately Constructed Systemic Representations of the World of Activity to Dialogue: A Five-Level Communication Architecture for Strong Language-Based ASI
arXiv cs.AI
arXiv:2610.32869v1 Announce Type: new Abstract: This paper proposes a five-level organization of message formation and understanding within thinking and activity in which ASI participates. Its central thesis is that a language model's…
-
Prediction-Powered Data Fusion for Treatment Effect Estimation
arXiv cs.LG
arXiv:2610.12332v2 Announce Type: replace-cross Abstract: Randomized controlled trials (RCTs) identify treatment effects without confounding but are often small, whereas observational studies (OBS) are large but may be confounded. Many…