Research

Latest Research

A running feed of newly published AI research papers, pulled daily from arXiv and filtered by hand before it appears here.

Papers

50 items
Keyword
01VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding HarnessarXiv cs.AIarXiv:2610.19424v2 Announce Type: replace-cross Abstract: LLM system capabilities depend not only on model parameters but also on the executable harnesses that organize information acquisition, tool use, reasoning, and memory. Recursive harness evolution can automate their design, but how it can reliably discover generalizable systems from limited, noisy, and repeatedly reused feedback remains unclear. We study this question in agentic video understanding, where rich multimodal evidence and cou19 Oct 2026↗02Towards Agentic Cloud Engineering: Graph and Loop Engineering with a Zero-Trust Agent HarnessarXiv cs.AIarXiv:2609.00050v2 Announce Type: replace-cross Abstract: Agentic AI is enabling cloud-based workflows in which autonomous agents reason over operational state, invoke authorized tools, modify software and infrastructure, deploy services, verify execution outcomes, and adapt across long-horizon, multistep tasks. Engineering such workflows requires explicit mechanisms for workflow progression, constrained execution, failure recovery, and verifiable completion. We present Agentic Cloud Workflow E19 Oct 2026↗03IndicDetect: Evaluating Cross-Lingual LLM-Generated Text Detection for Hindi, Telugu, and TamilarXiv cs.AIarXiv:2608.29919v2 Announce Type: replace-cross Abstract: The rapid proliferation of LLMs has further heightened the need to develop dependable AI-generated text detection, especially beyond English. Nevertheless, current benchmarks pay little attention to Indic languages and test detectors in idealized settings that do not represent the real world. We present a generalized benchmark for AI-generated text detection in Hindi, Telugu, and Tamil, which we call IndicDetect, designed to assess the r19 Oct 2026↗04BenthicDINO: Physics-Informed Self-Distillation for View-Invariant Side-Scan Sonar RepresentationsarXiv cs.AIarXiv:2608.23215v2 Announce Type: replace-cross Abstract: Automated interpretation of side-scan sonar (SSS) imagery is traditionally hindered by view-dependence, as recorded intensities inextricably conflate intrinsic seabed reflectivity with transient acquisition geometries and acoustic artifacts. To address this, we introduce BenthicDINO, a physics-informed self-supervised learning framework that decouples acquisition dynamics from seabed representations. Utilizing a ConvNeXt-v2 backbone, phy19 Oct 2026↗05ADEPT: Accelerating Dexterity via Pre-Training and Post-Training using Reinforcement LearningarXiv cs.AIarXiv:2608.19182v2 Announce Type: replace-cross Abstract: We introduce Accelerating Dexterity via Pre-Training (ADEPT), a large-scale reinforcement learning (RL) framework for learning sim-to-real transferable dexterity across high degree-of-freedom (DoF) robot embodiments that can solve long-horizon tasks directly from raw visuo-tactile perception. ADEPT pretrains a dexterous policy on a generic object reposing task, then post-trains downstream policies with this pretrained behavior as a prior19 Oct 2026↗06An AI Approach to Verified Production Cryptographic LibrariesarXiv cs.AIarXiv:2608.00965v2 Announce Type: replace-cross Abstract: Cryptographic code is critical infrastructure that must be correct, yet formally verifying production libraries remains difficult. Existing language-model proof systems solve isolated obligations with specifications and premises already given, leaving production-library verification unresolved. We present CryptoProver, an AI-based system that synthesizes internal specifications and Verus-checked proofs from high-level API contracts. Wi19 Oct 2026↗07Efficient LLM-Generated Shuttling Compilers for Complex Trapped-Ion ArchitecturesarXiv cs.AIarXiv:2607.24714v2 Announce Type: replace-cross Abstract: Trapped-ion quantum computers rely on shuttling compilers, which cast an input algorithm into a sequence of ion-qubit movements within a given architecture. We present the first study in which a single frontier large language model (LLM), Claude Opus 4.7, generates and iteratively refines the Python code of shuttling compilers from written specifications. Starting with a compiler for (i) a linear segmented trap, we extend it to (ii) a tr19 Oct 2026↗08KAYROS: An Anytime and Exact Open-Source Solver for Duration-Minimization Time-Dependent Vehicle Routing. A Technical Report and a Case Study in Human-AI EngineeringarXiv cs.AIarXiv:2607.23116v3 Announce Type: replace-cross Abstract: Time-dependent routing recognizes that the same journey can take a different time depending on when it begins. Under duration minimization, even the departure time of each vehicle becomes a decision. Exact methods for this setting exist in the literature, but researchers and practitioners have lacked a ready-to-use open solver that combines rich piecewise-linear travel times, early feasible solutions and optimality claims. KAYROS fills t19 Oct 2026↗09Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber TasksarXiv cs.AIarXiv:2607.21763v2 Announce Type: replace-cross Abstract: Large language model (LLM) agents routinely cheat on cybersecurity benchmarks, inflating reported pass rates far beyond genuine capability. Prior audits of Cybench found cheating in 0.3-3.4% of traces, implicating only a handful of models. We present a controlled prompt-ablation study across 22 frontier models from 7 providers on 23 Cybench capture-the-flag (CTF) challenges under three prompt conditions (no anti-cheat, standard, severe).19 Oct 2026↗10How Agents Ask for Permission: User Permissions for AI Agents, from Interfaces to EnforcementarXiv cs.AIarXiv:2607.13718v3 Announce Type: replace-cross Abstract: As AI agents grow in prevalence, so too do their risks to users, creating a need for robust and usable systems for agent permissions. Many recent proposals implement product-level permissions, where the same security policy is applied for all users of a given agent framework. Yet different users have varying needs, necessitating support for user-level permission policies in agentic AI systems. We survey 21 proposed user-level agent per19 Oct 2026↗11Do Multi-Agent LLMs Explore Each Other?arXiv cs.AIarXiv:2607.11250v2 Announce Type: replace-cross Abstract: Exploration is essential for reliable autonomy in multi-agent systems, yet it remains unclear whether large language model (LLM) agents can explore effectively when interacting with one another. We show that modern LLM agents often fail to do so, exhibiting myopic and polarized interaction patterns that lead to suboptimal coordination and increased regret. We formalize this challenge as the Multi-Agent Exploration problem, modeling it as19 Oct 2026↗12Is Your NPU Ready for LLMs? Dissecting the Hidden Efficiency Bottlenecks in Mobile LLM InferencearXiv cs.AIarXiv:2607.05475v2 Announce Type: replace-cross Abstract: Deploying Large Language Models (LLMs) on mobile devices enhances privacy and reduces latency, but is severely constrained by hardware inefficiencies. We present the first comprehensive, cross-layer measurement study of mobile LLM inference, spanning five mainstream frameworks (e.g., llama.cpp and GENIE) and three hardware backends (CPU, GPU, and NPU). To enable this analysis, we develop PowerBench, a fine-grained profiling tool that mea19 Oct 2026↗13Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement LearningarXiv cs.AIarXiv:2606.22995v2 Announce Type: replace-cross Abstract: Group-based Reinforcement Learning (RL) has significantly enhanced Large Language Models (LLMs) in agentic scenarios. To achieve finer-grained policy updates, recent agentic RL frameworks have shifted from trajectory-level to step-level training. However, long-horizon agentic RL suffers from severe reward sparsity and delay, as feedback is often deferred for dozens of interaction steps. While existing step-level frameworks refine trainin19 Oct 2026↗14Beyond Tokenization: Direct Timestep Embedding and Contrastive Alignment for Time-Series Question AnsweringarXiv cs.AIarXiv:2606.18986v2 Announce Type: replace-cross Abstract: Recent advances in large language models (LLMs) have given rise to time-series question answering (TSQA), which formulates time-series analysis as natural-language question answering. However, directly feeding raw numerical series into LLMs suffers from a tokenization bottleneck: Byte Pair Encoding fragments continuous values into unstable tokens whose embeddings lack meaningful metric structure, resulting in the loss of magnitude, scale19 Oct 2026↗15Accounting for Purchased Artificial Intelligence Inference in Corporate Greenhouse Gas Inventories: A Token-Based MethodarXiv cs.AIarXiv:2606.10660v2 Announce Type: replace-cross Abstract: Purchased artificial intelligence (AI) inference belongs in Scope 3 Category 1 of a corporate greenhouse gas inventory, yet practice omits it or applies a spend-based factor that gives 1.7 to 12.3 times the physical estimate and cannot tell one model or region from another. This paper sets out a token-based method and open dataset: a cradle-to-gate boundary, four tiers of activity data, an effective-token basis, a four-term factor (servi19 Oct 2026↗16CauTion: Knowing When to Trust LLMs for Ensemble Causal DiscoveryarXiv cs.AIarXiv:2606.03602v2 Announce Type: replace-cross Abstract: Causal discovery from observational data remains challenging due to the fundamental limitations of purely statistical methods, such as statistical distinguishability within equivalence classes and sensitivity to finite sample sizes. While large language models (LLMs) offer a promising source of domain knowledge to complement statistical inference, existing LLM-augmented methods are vulnerable to LLM errors and incur high token costs. Mor19 Oct 2026↗17SciIntBench: Measuring LLM Compliance with Research Integrity Norms Under Adversarial FramingarXiv cs.AIarXiv:2605.29468v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly used to support scientific work, but it is unclear whether they uphold responsible conduct of research (RCR) norms or help undermine them. We introduce SciIntBench, an adversarial benchmark of 1350 prompts across 270 scenarios, ten RCR categories, and three scientific domains. Each scenario appears in five gradations, from Most Overt to Most Covert adversarial prompts, and an additional Benig19 Oct 2026↗18CIDR: A Large-Scale Industrial Source Code Dataset for Software Engineering ResearcharXiv cs.AIarXiv:2605.12153v3 Announce Type: replace-cross Abstract: We present the Curated Industrial Developer Repository (CIDR), a large-scale dataset of real-world software repositories collected from industrial partners. The dataset comprises 4,687 repositories spanning 84 programming languages, totaling 1,033 million raw lines of code (670,3 million logical lines), along with structured metadata at the repository level, full version control history, and engineering-practice attributes such as contin19 Oct 2026↗19STRIDE: Reasoning-Aware Training for Time Series ForecastingarXiv cs.AIarXiv:2605.08625v2 Announce Type: replace-cross Abstract: Time Series Foundation Models (TSFMs) excel at numerical forecasting but operate as black boxes lacking qualitative reasoning. Conversely, applying LLMs directly to temporal data introduces a modality gap: text tokenizers fragment continuous numerical values, degrading mathematical relationships and exploding sequence lengths, leading to computational overhead. To resolve this, we introduce STRIDE (Strategic Time-series Reasoning Injecte19 Oct 2026↗20RamanBench: A Large-Scale Benchmark for Machine Learning on Raman SpectroscopyarXiv cs.AIarXiv:2605.02003v4 Announce Type: replace-cross Abstract: Machine learning (ML) has transformed many scientific fields, yet key applications still lack standardized benchmarks. Raman spectroscopy, a widely used technique for non-invasive molecular analysis, is one such field where progress is limited by fragmented datasets, inconsistent evaluation, and models that fail to capture the structure of spectral data. We introduce RamanBench, the first large-scale, fully reproducible benchmark for ML 19 Oct 2026↗