11718 papers
This paper introduces CCR-Bench, a novel benchmark designed to rigorously evaluate large language models on complex, real-world industrial tasks involving entangled content-format requirements and intricate logical workflows, revealing significant performance gaps in even state-of-the-art models.
This paper introduces a particle filtering framework to rigorously analyze the accuracy-cost tradeoffs of parallel inference methods in large language models, establishing theoretical guarantees and identifying fundamental limits while demonstrating that sampling error alone does not fully predict final model accuracy.
This paper introduces VLM-SubtleBench, a comprehensive benchmark spanning ten fine-grained difference types across diverse domains like industrial, medical, and aerial imagery, to evaluate and reveal the significant performance gaps between current vision-language models and humans in subtle comparative reasoning tasks.
This paper proposes using image segmentation as a visual diagnostic framework for hedonic games, demonstrating how granularization parameters influence coalition equilibrium structures and their ability to recover foreground ground-truth on the Weizmann benchmark.
This paper proposes a new approach that leverages LaCAM*'s ability to construct a dynamic, lightweight traffic map during search, overcoming the computational overhead and static limitations of existing Frank-Wolfe-based guidance path methods to achieve higher solution quality in Multi-Agent Path Finding.
This paper presents a decision-theory framework and a blended AI-statistical forecasting system that successfully delivered skillful, tailored monsoon onset predictions to 38 million Indian farmers in 2025, enabling better agricultural decision-making under uncertainty.
This paper introduces SMGI, a structural theory of general artificial intelligence that formalizes learning as the controlled evolution of a typed meta-model, unifying diverse existing approaches under a rigorous framework defined by structural closure, dynamical stability, bounded capacity, and evaluative invariance.
EveryQuery is a novel electronic health record foundation model that achieves efficient, zero-shot clinical prediction by directly estimating outcome likelihoods through task-conditioned pre-training, thereby outperforming computationally expensive autoregressive baselines—particularly for rare events—while currently facing limitations in complex disjunctive reasoning tasks.
This paper presents a vision-only autonomous bronchoscopy framework utilizing hierarchical long-short agents and a world-model critic to achieve accurate, sensor-free intraoperative navigation in preclinical models, demonstrating performance comparable to expert human operators.
The paper introduces Ares, a framework that dynamically selects the optimal reasoning effort level for each step of an LLM agent's task using a lightweight router, achieving up to 52.7% reduction in token usage with minimal impact on success rates compared to static high-effort strategies.
This paper introduces Rel-MOSS, a novel relation-centric deep learning framework that addresses the critical issue of class imbalance in relational databases by employing a relation-wise gating controller and a relation-guided minority synthesizer to enhance the representation and over-sampling of minority entities, thereby significantly outperforming existing methods in entity classification tasks.
The paper proposes IMSE, a test-time adaptation method that fine-tunes only the singular values of Vision Transformer linear layers via a spectral mixture of experts and a diversity maximization loss to prevent feature collapse, achieving state-of-the-art performance with significantly fewer trainable parameters.
SWE-Fuse is a novel training framework that enhances software engineering agents by fusing issue-free trajectory learning with entropy-aware RLVR to overcome the limitations of noisy real-world issue descriptions, achieving state-of-the-art performance on the SWE-bench Verified benchmark.
This paper proposes that intelligent, human-supervised AI agents built on deep learning and large language models represent the next evolution of the scientific method, enabling researchers to manage unprecedented data complexity and scale discovery, as demonstrated by the Dr. Sai system in particle physics.
This paper introduces ELLMob, a self-aligned Large Language Model framework that leverages Fuzzy-Trace Theory to reconcile habitual patterns with event constraints, addressing the lack of event-annotated datasets and significantly improving the generation of human mobility trajectories during major societal events like typhoons, pandemics, and the Olympics.
This paper introduces PSTNet, a lightweight, physics-structured neural network with only 552 parameters that embeds atmospheric turbulence scaling laws directly into its architecture to enable accurate, real-time turbulence estimation on resource-constrained aircraft guidance systems where traditional models fail.
This paper introduces EvoStage, a novel evolutionary paradigm that leverages large language models with a stagewise, multi-agent approach and real-time feedback to overcome the limitations of black-box modeling, successfully generating algorithm designs that outperform both human experts and existing methods in complex industrial tasks like chip placement and black-box optimization.
This paper introduces HILA, a Human-In-the-Loop Multi-Agent Collaboration framework that employs Dual-Loop Policy Optimization to train agents with metacognitive policies for dynamically deferring to human experts and continuously improving their reasoning capabilities, thereby overcoming the static knowledge limitations of purely autonomous systems.
VORL-EXPLORE is a hybrid learning and planning framework for multi-robot exploration in dynamic environments that couples task allocation with motion execution via a shared navigability fidelity signal, enabling adaptive arbitration between global and reactive policies to prevent bottlenecks and ensure robust, collision-free coverage.
The paper introduces OSExpert, a computer-use agent that leverages a GUI-based depth-first search exploration algorithm to discover action primitives and self-construct a skill curriculum, thereby significantly improving performance and efficiency on complex tasks to approach human expert levels.