Tool use, function calling, orchestration and the protocols that let models act rather than only answer.
arXiv:2602.03318v4 Announce Type: replace Abstract: Operations Research (OR) relies on expert-driven modeling--a slow and fragile process ill-suited to novel scenarios. While large language models (L...
arXiv:2608.22160v1 Announce Type: new Abstract: Physical automation is scaling toward fleets of embodied machines commanded by an AI brain. Early deployments already run factories and warehouses at p...
arXiv:2608.11596v1 Announce Type: cross Abstract: Modern radar systems require adapting their processing strategies in response to changing interference, clutter, and data availability. This paper in...
arXiv:2608.23030v1 Announce Type: new Abstract: We study how an AI system can identify and model other agents in its environment from observation alone, which is a capability necessary for cooperativ...
arXiv:2608.22723v1 Announce Type: new Abstract: This workshop paper comprehensively reviews the First Challenge on Unified Removal of Raindrops and Reflections. The challenge aims to address a freque...
arXiv:2608.21377v1 Announce Type: cross Abstract: Sycophancy in large language models, the tendency to prioritize user agreement over truthful responses, has been documented extensively but studied p...
VeCAS is a two‑stage framework that generates contrast‑free X‑ray angiograms from non‑contrast images. Stage I localizes vascular structures using a discriminative model and cross‑modality latent distillation, while Stage II synthesizes angiographic appearance within those regions with a vessel‑focused inpainting model. Experiments on a lower‑limb vascular intervention dataset show that VeCAS improves vascular structural fidelity and image quality, and robotic guidewire navigation tests demonstrate significant reductions in time to target and operation steps.
The paper introduces ADN‑Agent, an architecture that uses a large language model to orchestrate multiple domain‑specific models (DSMs) for active distribution network (ADN) management. It features adaptive intent recognition, task decomposition, and a unified communication interface for heterogeneous DSMs, along with a pipeline for fine‑tuning small language models on language‑intensive subtasks. Experiments show ADN‑Agent outperforms existing LLM application paradigms in coordinating DSMs for complex ADN operations.
TO-Agents is a multi‑agent AI framework that translates natural‑language design intent into iterative topology optimization. It converts a human problem description into solver inputs, runs the optimizer, renders 3D topologies, and employs a judge agent to critique and revise results using multiview vision‑language reasoning. Evaluated on a cantilever beam and a phone‑stand design, the system achieved preference‑aligned designs in 60% of trials, outperforming an ablated pipeline by up to six times and enabling end‑to‑end intent‑to‑prototype design with additive manufacturing.
The paper introduces HOT-POT, a method that applies optimal transport to sparse stereo matching, addressing challenges such as occlusions, motion, and camera distortions. By modeling camera‑projected points as half‑lines and using epipolar and 3D ray distances as cost functions, the authors formulate efficient assignment problems for unsupervised sparse matching. The approach is extended to hierarchical optimal transport for unsupervised object matching, with experiments demonstrating its effectiveness in facial analysis for aligning different landmarking conventions.
arXiv:2608.21836v1 Announce Type: new Abstract: Large language models have become increasingly capable agents for low-level code and kernel optimization, but isolated kernel benchmarks provide only a...
ViSMoE introduces a visual‑aware sparse Mixture‑of‑Experts framework for embodied referring expression grounding, enabling an agent to navigate real environments and localize a target object from natural language instructions. By routing visual information through specialized experts, the method produces discriminative representations for both navigation views and candidate objects, unlike prior approaches that use a single vision encoder. Experiments on the REVERIE and SOON datasets show that ViSMoE surpasses existing state‑of‑the‑art methods.
DRAgent introduces a discriminative reasoning framework for Referring Expression Segmentation that replaces direct coordinate prediction with a two‑stage selection process. First, a detector generates candidate boxes; then a multimodal large language model (MLLM) discriminates the correct target through high‑recall screening and instance‑wise verification. The chosen box serves as a spatial prompt for a foundation segmentation model, and a self‑consistency‑filtered data pipeline is used for LoRA‑based fine‑tuning, yielding competitive results on RefCOCO, RefCOCO+, and RefCOCOg.
The paper proposes a new way to describe benchmarks for AI systems that perform knowledge work, outlining four explicit fields: represented activity, tested setting, required work product, and evaluated result. It builds an inventory of 18 work activities from O*NET to enable activity-level reporting across occupations, and evaluates these activities for semantic coherence, algorithm sensitivity, ontology legibility, and human interpretability. The authors apply their framework to three existing benchmarks—GDPval, OfficeQA Pro, and APEX-SWE—to show how different aspects of work can be captured within the same representation.
Chain of Operators (CHOP) is a framework that enables frozen scientific foundation models to tackle out‑of‑distribution (OOD) tasks without any parameter updates. By leveraging In‑Context Operator Networks (ICON), CHOP breaks down unfamiliar problems into sequences of explicit mathematical operations and model calls, effectively translating OOD queries into the model’s learned regime. Experiments on PDE benchmarks and air‑quality forecasting show that CHOP consistently reduces inference errors while maintaining full interpretability and cross‑equation generalization.
The paper "SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding Agents" investigates how agent skills—task‑specific instructions, scripts, and resources—can be exploited to create a trusted instruction channel that enables token amplification attacks. It introduces a two‑phase framework, SkillBloat, which first screens a library of attack‑type conditions across multiple amplification mechanisms and then refines the strongest candidate through LLM‑guided full‑document skill rewriting. Evaluated on a real‑world skill benchmark, SkillBloat achieves an average best amplification of 5.4184×–10.1455× across multiple coding‑agent target configurations, and an ablation study shows that the second‑stage refinement consistently improves performance over the initial screening alone.
arXiv:2601.19066v3 Announce Type: replace-cross Abstract: Bug Reproduction Tests (BRTs) have been used in many Automated Program Repair (APR) systems, primarily for validating fixes and aiding fix ge...
The study examines how Theory of Mind (ToM) reasoning and prosocial beliefs influence large language models (LLMs) in the ultimatum game. By initializing LLM agents with Greedy, Fair, or Selfless beliefs and applying chain‑of‑thought or varying levels of ToM reasoning, the authors ran 2,700 simulations across several models, including o3‑mini and DeepSeek‑R1 Distilled Qwen 32B. Results show that ToM‑enhanced LLMs align more closely with human decision patterns, exhibit greater consistency, and achieve better negotiation outcomes, with Llama 3.3 70B producing the most belief‑consistent reasoning. whyItMatters":"The findings clarify the importance of incorporating Theory of Mind into LLMs to improve their alignment with human norms in cooperative decision‑making tasks."
The paper introduces an agentic AI Scientist workflow that automates the entire baseline development process for medical imaging by combining literature-guided reasoning, automated code generation, and hypothesis-driven experimentation. Evaluated on four public benchmarks covering segmentation, classification, and detection, the pipeline consistently improves validation performance, achieving competitive leaderboard results such as 6th place on both PUMA tracks and 31st on MILK10k. The approach also shows strong domain generalization on MIDOG25 across scanners, tumor types, and species, demonstrating that a skill-based, literature-guided agentic workflow can reduce engineering effort without task-specific redesign.
MPCFormer is a physics‑informed, data‑driven framework that explicitly models multi‑vehicle social interaction dynamics for autonomous driving. It uses a Transformer‑based encoder‑decoder to learn discrete state‑space dynamics from naturalistic data, enabling explainable, human‑like behavior planning within a Model Predictive Control (MPC) framework. In open‑loop NGSIM tests, it achieves the lowest trajectory prediction errors (ADE 0.86 m over 5 s), and in closed‑loop intense interaction scenarios it attains a 94.67 % planning success rate, 15.75 % efficiency gain, and reduces collisions from 21.25 % to 0.5 %.