arXiv subjects

Jiayu Wu

Publications and source records attributed to Jiayu Wu.

7 recordsLinked to original sources

arXiv2026

Interactor: Agentic RL oriented Iterative Creation for Ad Description Generation in Sponsored Search

This paper focuses on automatically generating informative ad descriptions in sponsored search. Unlike ad titles which are usually optimized to attract user click feedbacks, ad descriptions have a longer text span and possess the potential of incorporating world knowledge to address user search intents while presenting the fine-grained selling points of the ads. We propose Interactor, a multi-turn iterative creation framework optimized with agentic RL for ad description generation. The generation model acts as a policy that interacts with a customized environment consisting of multiple generative reward models. Given initial generations by the policy, the customized GenRMs evaluate multi-dimensional qualities including knowledge capacity and landing page consistency, providing both binary signals and reasoning feedbacks. The policy then iteratively refines the descriptions based on such feedbacks to ensure continuous improvement. Experiments on industrial datasets show that the Interactor framework significantly outperforms state-of-the-art approaches in generating knowledge-rich and faithful ad descriptions. Since May 2026, it has been deployed online in a leading search ads system, contributing to both ad revenue and user experience.

cs.CL↗

arXiv2025

Data-driven functional state estimation of complex networks

The internal state of a dynamical system, a set of variables that defines its evolving configuration, is often hidden and cannot be fully measured, posing a central challenge for real-time monitoring and control. While observers are designed to estimate these latent states from sensor outputs, their classical designs rely on precise system models, which are often unattainable for complex network systems. Here, we introduce a data-driven framework for estimating a targeted set of state variables, known as functional observers, without identifying the model parameters. We establish a fundamental functional observability criterion based on historical trajectories that guarantees the existence of such observers. We then develop methods to construct observers using either input-output data or partial state data. These observers match or exceed the performance of model-based counterparts while remaining applicable even to unobservable systems. The framework incorporates noise mitigation and can be easily extended to nonlinear networks via Koopman embeddings. We demonstrate its broad utility through applications including sensor fault detection in water networks, load-frequency control in power grids, and target estimation in nonlinear neuronal systems. Our work provides a practical route for real-time target state inference in complex systems where models are unavailable.

eess.SY↗

arXiv2025

Matrix representation and GPU-optimized parallel B-spline computing

B-spline modeling is fundamental to CAD systems, and its evaluation and manipulation algorithms currently in use were developed decades ago, specifically for CPU architectures. While remaining effective for many applications, these algorithms become increasingly inadequate as CAD models grow more complex, such as large-scale assemblies and microstructures. GPU acceleration offers a promising solution, but most existing GPU B-spline algorithms simply adapt CPU counterparts without accounting for the mismatch between the unstructured, recursive nature of B-splines and the structured nature of GPU kernels, ultimately failing to fully leverage GPU capabilities. This paper presents a novel approach that transforms B-spline representations into regular matrix structures, reducing all evaluation and manipulation computations to matrix addition and multiplication, thus better aligning with GPU architecture. By combining this matrix representation with GPU-optimized task scheduling and memory access patterns, the paper demonstrates significant performance improvements in the key B-spline operations of inversion and projection. Experimental results show an improvement of about two orders of magnitude in computational speed compared to existing methods.

cs.DC↗

arXiv2025

Flow-Aware Navigation of Magnetic Micro-Robots in Complex Fluids via PINN-Based Prediction

While magnetic micro-robots have demonstrated significant potential across various applications, including drug delivery and microsurgery, the open issue of precise navigation and control in complex fluid environments is crucial for in vivo implementation. This paper introduces a novel flow-aware navigation and control strategy for magnetic micro-robots that explicitly accounts for the impact of fluid flow on their movement. First, the proposed method employs a Physics-Informed U-Net (PI-UNet) to refine the numerically predicted fluid velocity using local observations. Then, the predicted velocity is incorporated in a flow-aware A* path planning algorithm, ensuring efficient navigation while mitigating flow-induced disturbances. Finally, a control scheme is developed to compensate for the predicted fluid velocity, thereby optimizing the micro-robot's performance. A series of simulation studies and real-world experiments are conducted to validate the efficacy of the proposed approach. This method enhances both planning accuracy and control precision, expanding the potential applications of magnetic micro-robots in fluid-affected environments typical of many medical scenarios.

cs.RO↗

arXiv2024

Inductive-Deductive Strategy Reuse for Multi-Turn Instructional Dialogues

Aligning large language models (LLMs) with human expectations requires high-quality instructional dialogues, which usually require instructions that are diverse and in-depth. Existing methods leverage two LLMs to interact for automatic collection: one simulating a user to pose instructions, and the other acting as a system agent to respond. However, these user simulators struggle to model the rules behind how dialogues can pose different instructions without explicit guidance, resulting in general instructions. In this paper, we propose to explicitly capture the complex rules to help the user simulator pose diverse and in-depth instruction. Specifically, we first induce high-level instruction strategies from various real instruction dialogues serving as rules. Afterward, different possible strategies are applied to the newly given dialogue scenario deductively to pose various instructions. Experimental results show that our method can generate diverse and in-depth instructions. The constructed multi-turn instructional dialogues can outperform competitive baselines on the downstream chat model.

cs.CL↗

arXiv2020

Inclusive diffractive heavy quarkonium photoproduction including quark subprocesses

The inclusive $J/Ψ$, $Ψ(2S)$ and $Υ(1S)$ direct and resolved photoproduction are investigated by including the quark subprocesses in the framework of non-relativistic quantum chromodynamics (NRQCD). We find that the theoretical total cross section of heavy quarkonium productions are in good agreement with the data available at HERA, once the $γq$, $qg$ and $qq$ subprocesses in the heavy quark pair productions are taken into account. The inclusive diffractive rapidity and transverse momentum distributions of $J/Ψ$, $Ψ(2S)$ and $Υ(1S)$ in $pp$, $pPb$ and $PbPb$ collisions at LHC are also studied by our quark improved NRQCD model combined with the resolved pomeron model. We find that the contributions from the quark involved subprocesses can reach to $8\%$ in the rapidity distribution and $6\%$ in the transverse momentum distribution. The numerical results show that the contributions from quark involved subprocesses are significant in heavy quarkonium photoprodution.

hep-ph↗

arXiv2020

Theory, preparation, properties and catalysis application in 2D Graphynes-Based Materials

Carbon has three hybridization forms of sp-, sp2- and sp3-, and the combination of different forms can obtain different kinds of carbon allotropes, such as diamond, carbon nanotubes, fullerene, graphynes (GYs) and graphdiyne (GDY). Among them, the GDY molecule is a single-layer two-dimensional (2D) planar structure material with highly -conjugation formed by sp- and sp2- hybridization. GDY has a carbon atom ring composed of benzene ring and acetylene, which makes GDY have a uniformly distributed pore structure. In addition, GDY planar material have some slight wrinkles, which makes GDY have better self-stability than other 2D planar materials. The excellent properties of GDY make it attract the attention of researcher. Therefore, GDY is widely used in chemical catalysis, electronics, communications, clean energy and composite materials. This paper summarizes the recent progress of GDY research, including structure, preparation, properties and application of GDY in the field of catalysts.

cond-mat.mtrl-sci↗