📄 Notable* Recent AI/ML arXiv Papers

Last updated just now...

📄 Mapping the AI Supply Chain: Dependencies, Vulnerabilities, and Market Concentration in the AI Ecosystem
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24927v1
👥 Authors: Sarah H. Cen (possible past Deepmind (United Kingdom) affiliation), Lindsey A. Gailmard, Joohan Kim, Rishi Bommasani, Daniel E. Ho, Percy Liang (possible past Stanford University affiliation)
Abstract

AI systems are built across many organizations that contribute models, data, compute, capital, talent, and other assets. While the layers of the AI stack are relatively well understood, the relationships between actors and components across layers remains sparsely documented. In this work, we focus on the organizations that make up the AI supply chain: the complex network of actors that contribute to the development and deployment of AI. We motivate the examination of the organizational AI suppl...

📄 WorldRuler: Interactive World Modeling under Configurable Rules
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24869v1
👥 Authors: Mingliang Zhai, Chuanhao Li, Jiaming Tan, Zhen Li (possible past Google (United States) affiliation), Yuwei Wu, Yunde Jia, Zian Meng, Yuanyang Yin, Kaipeng Zhang (possible past Tencent (China) affiliation)
Abstract

Rules govern how player actions change interactive worlds. Different rules create distinct gameplay experiences within similar scenes and action spaces. For customizable gameplay, generative world models must follow user-defined rules beyond real-world behaviors. However, realizing user-defined rules presents two challenges: Rule-State Persistence and Rule-to-Visual Realization. The former requires preserving rule definitions, dependencies, and world states across saving and loading, so that res...

📄 Piccolo: Co-Evolving Decision Models and Data through Recursive Self-Improvement
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24651v1
👥 Authors: Zhenqing Ling, Qirui Jiao, Daoyuan Chen, Cheng Chen (possible past Google (United States) affiliation), Xinji Mai, Haowen Gao, Zhuo Zhang, Xuchen Pan, Mengsu Ding, Jiaji Deng, Yuexiang Xie, Xiaoming Qin, Weirui Kuang, Zhaoyang Liu, Yaliang Li (possible past Baidu (China) affiliation), Bolin Ding, Jingren Zhou
Abstract

We present Piccolo, a family of multimodal decision models (0.8B to 27B) that coevolves model capabilities and training data via Agentic Data Engineering and recursive self-improvement (RSI). The models predict probabilities over candidate answers for general-purpose text and visual judgments, agent harnesses, data workflows, and interactive games. Piccolo-27B tops the Decision Index Vision Full leaderboard with a score of 75.59, while Piccolo-9B and Piccolo-4B achieve 68.35 and 67.46, respectiv...

📄 H3Remover: Reinforcing Video Object Removal without Paired Data
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24633v1
👥 Authors: Kangjie Chen, Xiangyu Li, Jinhao Deng, Dongbin Zhang, Chaoda Zheng, Shijia Chen, Yuqi Zhang, Choo Sin Wai, Boyi Sun, Puyi Wang, Yu Zhang (possible past Google (United States) affiliation), Xianming Liu (possible past Meta (United States) affiliation), Boyang Wang
Abstract

Pretrained video editors possess rich removal and completion priors, yet may retain targets or alter unrelated content. We present H3Remover, which reinforces MiniMax-H3 Base Ref2VA for video object removal without paired object-free training targets. To discourage superficial edits such as recoloring, a target-removal reward favors calibrated depth changes consistent with revealing farther content. To limit unintended scene changes, a background-geometry reward encourages preservation of relati...

📄 BaseRSI: Recursive Self-Improvement of a Base Vision-Language Model via Training-Free Harness Internalization
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24578v1
👥 Authors: Mingda Zhang, Yuanhao Dong, Qiang Huang, Rui Qian (possible past Shanghai Jiao Tong University affiliation), Yanjin Li, Zhenyu Liu (possible past Google (United States) affiliation), Xiao Huang, Zihao Yang, Jian Wang (possible past Baidu (China) affiliation), Pinjia He, Xiaoying Tang
Abstract

Recursive self-improvement (RSI) is moving from the harness of prompts and tools around a base model toward the model itself. However, existing RSI still faces key challenges, including harness-bound gains that stay outside the base model, black-box supervision that reads only the model's outputs, and imprecise edits whose site and strength are fixed in advance. To address these challenges, we propose BaseRSI, a new paradigm of Recursive Harness Internalization: the base model improves its harne...

📄 OpenKAGE: Open-Ended Lifelong Learning in the Open World via Knowledge-Sharing Agent-Group Evolution
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24302v1
👥 Authors: Wenhao Yuan, Longhui Yuan, Zexi Jia, Ningzhe Shi, Jingwei Li, Shipeng Cen, Pengcheng Luo, Jie Zhou (possible past Tsinghua University affiliation), Jinchao Zhang (possible past Tencent (China) affiliation)
Abstract

Building autonomous agents that continually explore open worlds and acquire new capabilities through experience is a long-standing aspiration in artificial intelligence. Existing research often pursues lifelong learning within predefined tasks, curricula, or objectives, with self-improvement largely confined to individual agents. We introduce OpenKAGE, an agent-group evolution framework for open-ended lifelong learning. Within this framework, agents autonomously explore open worlds and continual...

📄 Worth Another Look:Constrained World-Model Planning for Visual Inspection
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24187v1
👥 Authors: Ziyue Kang, Zixi Qi, Yuchen Dai, Nana Lv, Xirong Cheng, Liang Liu (possible past Tencent (China) affiliation), Ying Li (possible past Meta (United States) affiliation), Haoyan Wei, Haiyang Yao
Abstract

Visual inspection in field robotics and environmental monitoring must localize multiple targets under strict compute and sensing caps. Exhaustive sweeps waste calls on absent classes, while myopic and search-free policies stumble over heterogeneous tariff cliffs or discard viewport edits whose value emerges only after re-querying. We formulate inspection over frozen detectors as neuro-symbolic observation planning under partial observability, decoupling a compact recurrent world model (WM) from ...

📄 Embodied Graph: A Relational World Model for Long-Horizon Tasks
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24115v1
👥 Authors: Zongyuan Wu, Xin Wang (possible past University Of Edinburgh affiliation), Tongtong Feng, Yuan Meng, Zeyang Zhang, Ling Feng, Wenwu Zhu (possible past Tsinghua University affiliation)
Abstract

Long-horizon tasks in embodied and web environments involve extended decision sequences and delayed feedback, making accurate planning crucial for success. World models tackle this by learning latent dynamics for imagination-based planning. However, existing approaches merely rely on dense pixel-level observations, which forces the model to absorb redundant information while overlooking relations among task-relevant entities. Consequently, small errors may accumulate over long imagined rollouts ...

📄 APS: Adaptive Partitioning-Based Frame Selection for Long Video Understanding
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24079v1
👥 Authors: Junxiang Wen, Yinxi Zhao, Jiawei Chen (possible past Tencent (China) affiliation), Jingying Hu, Lijun Wang (possible past Eth Zurich affiliation), Zhen Qin, Shuiguang Deng
Abstract

Recent advances in large vision-language models (LVLMs) have significantly improved video understanding capabilities, yet processing long videos remains challenging due to the substantial computational cost of extensive visual inputs. Frame selection offers an efficient alternative, but many existing region-based methods rely on fixed or predefined temporal partitions, which may result in suboptimal temporal regions, compromising subsequent budget allocation and frame selection. In this work, we...

📄 How Complete Must an Environment Be to Train a Capable GUI Agent?
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23946v1
👥 Authors: Jingxing Wang, Teng Wang, Xingyu Lou, Zihan Guo, Jun Wang (possible past Tencent (China) affiliation), Weiwen Liu, Weinan Zhang (possible past Shanghai Jiao Tong University affiliation), Jianghao Lin
Abstract

Graphical user interface (GUI) agents are widely trained in executable GUI environments. To lower cost, build environments in batches, and reduce manual effort, each work trades away part of the environment through its construction choices. We describe these choices along three dimensions: reward coverage, the goal-reaching trajectories the verifier rewards; data coverage, the application data the agent is exposed to beyond what the task reads or writes; and feature coverage, the application fea...

📄 GLIDE: Scalable and Event-Aware Inductive Forecasting over Evolving Logistics Graphs
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23942v1
👥 Authors: Zhi Sheng, Chuyue Wang, Jiahao Ji, Yile Liang, Xuetao Ding, Jinghua Hao, Jie Feng (possible past Tsinghua University affiliation), Yong Li (possible past Tsinghua University affiliation)
Abstract

Accurate forecasting of order volume at scale-effect hotspots (SEHs), spatially coherent groups of logistics flows that serve as distributed optimization units for real-time many-to-one dispatch, is a prerequisite for efficient order pooling in on-demand food delivery (OFD) platforms such as Meituan. However, real-world logistics networks present unique challenges that render standard spatiotemporal prediction model ineffective. (i) Dynamic Topology: Unlike static road networks, logistics graphs...

📄 Cardiac MRI-informed electrocardiography enables biologically grounded cardiomyopathy detection
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23925v1
👥 Authors: Zhengyao Ding, Kewen Wang, Yujian Hu, Ziyu Li, Yiheng Mao, Yue Wang, Haitao Li, Dengnan Zhang, Xuanhong Liu, Yue Wang, Jing Wang (possible past Google (United States) affiliation), Yue Chen (possible past Google (United States) affiliation), Longbo Wang, Mengjia Chen, Chengchen Zhao, Xuesen Chu, Yubin Zhang, Mor Peleg, Judith Somekh, Mingxuan Yuan, Hui Yan, Yue Gu, Fan Yi, Zhilong Jia, Zhentao Tang, Xi Hu, Fei Wu (possible past Google (United States) affiliation), Hongkun Zhang, Ting Chen (possible past Google (United States) affiliation), Zhengxing Huang
Abstract

Precise myocardial phenotyping at population scale is limited by the cost and accessibility of cardiac magnetic resonance (CMR), whereas electrocardiography (ECG) is widely available but reflects electrical rather than structural and functional information. Here we develop MyoECG, a cross-modal representation learning framework that distils myocardial structural and functional knowledge from CMR into standard 12-lead ECGs, enabling imaging-informed cardiomyopathy detection and subtype classifica...

📄 Asking but Missing: Decision Coverage in Natural-Language Data Querying
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23864v1
👥 Authors: Hangle Hu, Chenyu Hou, Bin Cao (possible past Microsoft (United States) affiliation), Jing Fan, Wei Luo (possible past Baidu (China) affiliation)
Abstract

Large language models are increasingly used to translate natural language questions about structured data into executable programs. A fundamental challenge is recognizing when a question admits multiple computational interpretations and identifying the decisions needed to resolve them. We investigate whether language models recognize these unresolved decisions before generating code, or silently adopt assumptions that may lead to incorrect results. We refer to the latter behavior as Underspecifi...

📄 Thinking Fast, Anywhere: One System-1 for the Decision Points of Any Agent Workflow
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23751v1
👥 Authors: Rui Li (possible past Google (United States) affiliation), Liyang He, Zheng Zhang, Zhenya Huang, Qi Liu (possible past Tencent (China) affiliation)
Abstract

Autonomous LLM agents incur substantial cost and latency by invoking a large model (System-2) at every step, even when many intermediate actions choose among a small set of known candidates. Existing dual-system agents offload such decisions to a lightweight System-1 model, but rely on workflow-specific hooks authored per environment. We present AnyS1, a framework that turns human-authored procedural skills, whose instructions naturally specify where choices arise and which options apply, into a...

📄 ARISE: Practical Self-Evolving Memory for Accurate and Cost-Efficient Enterprise Agents
🗓️ Published: 11/17/2026
🔗 http://arxiv.org/abs/2611.23588v1
👥 Authors: Jiangyun Zhang, Zhou Tang, Kristen Surrao, Zhiyu Chen, Albert Guan, Tor Nitayanont, Yupei Zhang, Roopali Singh, Yunkai Xiao, Shreya Pandit, Julia Huang, Shayan Ali Akbar, Erwin Cornejo, Yuan Li (possible past Google (United States) affiliation), Yi Zhang (possible past Google (United States) affiliation)
Abstract

Enterprise LLM agents operate in dynamic environments where foundation models evolve rapidly, workflows update frequently, and deployments need to satisfy accuracy and cost requirements. We propose ARISE (Agent Refinement through Iterative Self-Evaluation), a practical memory architecture in which the agent's memory evolves from its own failures without updating model weights, tools, or workflow logic. ARISE uses the model itself or a teacher model to attribute the failures and propose correctio...

📄 BOS Summary: The First Token Matters
🗓️ Published: 11/17/2026
🔗 http://arxiv.org/abs/2611.23508v1
👥 Authors: Hui Zhu, Saleh Zare Zade, Rafi Ibn Sultan, Deng Pan (possible past Tencent (China) affiliation), Arun Ross (possible past Google (United States) affiliation), Dongxiao Zhu
Abstract

Large vision-language models (LVLMs) and language models (LLMs) often exhibit attention sinks, whereby particular positions, such as the beginning-of-sequence (BOS) token, receive a disproportionate share of attention. Under standard causal masking, however, the BOS query attends only to itself, so the representation that later tokens repeatedly access carries no information about the input. We introduce BOS Summary, a training-free inference-time method that repurposes this persistent pathway t...

📄 HardwareDx: A Foundational Graph Model System for Datacenter Machine Repairs
🗓️ Published: 11/17/2026
🔗 http://arxiv.org/abs/2611.23398v1
👥 Authors: Yuanlin Wen, Elan Markowitz, Zubo Gu, Sami Abu-El-Haija (possible past Stanford University affiliation), Bryan Perozzi (possible past Google (United States) affiliation), Mikhail Galkin, Ali Parviz, Aayush Gupta, Venkata Chivukula, Griffin Hadfield, Aravind Akella, Daniel J. Frame, Vahab Mirrokni (possible past Google (United States) affiliation), Amin Vahdat (possible past Google (United States) affiliation)
Abstract

The rapid scale-up of AI infrastructure has made datacenter reliability a cornerstone of the cloud business economy. Since hardware failures in hyper-scale AI clusters can stall multi-million dollar training workloads, the downtime of individual machines can cause outsized economic losses. Minimizing machine downtime, therefore, is more important than ever. To address this, we present Hardware Diagnoser (HardwareDx), an AI-driven hardware diagnosis system that applies a graph foundation model ...

📄 Ambient LLMs: Data-Efficient Supervised Fine-tuning via Temperature Matching
🗓️ Published: 11/17/2026
🔗 http://arxiv.org/abs/2611.23229v1
👥 Authors: Giannis Daras, Marios Papachristou, Adam Klivans, Antonio Torralba (possible past Massachusetts Institute Of Technology affiliation), Constantinos Daskalakis (possible past University Of California, Berkeley affiliation)
Abstract

Collecting large-scale, diverse, and high-quality post-training datasets for autoregressive language models is challenging. Consequently, Supervised Fine-Tuning (SFT) is typically applied to datasets significantly smaller than those used for pre-training, which leads to overfitting and the catastrophic forgetting of the tails of the data distribution. To prevent this collapse, we introduce Ambient SFT, a principled framework for leveraging imperfect predictions to augment the fine-tuning set. In...

📄 QuantAtlas-OM: Taming Neural Spectral Bias in Long-Horizon Spatiotemporal Ocean Dynamics
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24124v1
👥 Authors: Hao Wu (possible past Tencent (China) affiliation), Yuan Gao (possible past Tencent (China) affiliation), Tong Wu, Fan Xu, Fan Zhang, Jinhan Li, Kun Wang, Zhenyu Ma, Bryan Dai, Xiaomeng Huang, Qingsong Wen
Abstract

Autoregressive neural networks exhibit spectral bias in long-horizon spatiotemporal prediction. As rollouts progress, high-frequency details decay, blurring the predicted fields and destabilizing long-range forecasts. Ocean dynamics offer a particularly stringent test because kinetic energy is concentrated in mesoscale eddies and fronts spanning tens of kilometers. This degradation stems from coupling large-scale phase evolution and small-scale energy retention in a shared feature space. Evoluti...

📄 Rethinking Contrastive Targets in Radiology Language-Image Pretraining
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.24016v1
👥 Authors: Yunhe Gao, Ashwin Kumar, Jiaming Liu (possible past Baidu (China) affiliation), Chong Wang (possible past Google (United States) affiliation), Maya Varma, Fuying Wang, Stefania L. Moroianu, Louisa Fay, Sergios Gatidis, Akshay S. Chaudhari, Curtis Langlotz
Abstract

CLIP-style contrastive image-text pretraining is widely used in radiology data, where each image is paired with one report and every other report in the batch is treated as a negative. Under this recipe, two failures appear, neither of which is systematically addressed by existing contrastive learning theory: zero-shot evaluation peaks then degrades while training loss keeps decreasing, and training fails to learn from large batches. We hypothesize the cause is structural rather than radiology-s...

📄 HypoLoop: Revising Weight-Editing Plans from Scientific Feedback
🗓️ Published: 11/18/2026
🔗 http://arxiv.org/abs/2611.23718v1
👥 Authors: Mengran Li, Xichen Yan, Bo Li (possible past Tencent (China) affiliation), Chengyang Zhang, Jinfeng Xu, Zheng Lin, Zelin Zang, Zhen Lei (possible past Beijing Academy Of Artificial Intelligence affiliation), Jiebo Luo, Victor C. M. Leung, Stan Z. Li
Abstract

Scientific experiments continually reveal prediction errors. Effective updates depend on which parameters change and which errors drive the correction. We introduce HypoLoop, a family of native weight-editing methods that connects these choices to experimental feedback. Fixed, numerically adaptive, and LLM-guided policies produce editing plans; numerical solvers compute weight patches, and internal validation selects the deployed predictor. Successive observations and validation records support ...

*Notable papers are those with at least two authors from a "big" AI/ML lab.