With the rapid growth of Web-based academic publications, more and more papers are being published annually, making it increasingly difficult to find relevant prior work. Citation prediction aims to automatically suggest appropriate references, helping scholars navigate the expanding scientific literature. Here we present CiteRAG, the first comprehensive retrieval-augmented generation (RAG)-integrated benchmark for evaluating large language models on academic citation prediction, featuring a multi-level retrieval strategy, specialized retrievers, and generators. Our benchmark makes four core contributions: (1) We establish two instances of the citation prediction task with different granularity. Task 1 focuses on coarse-grained list-specific citation prediction, while Task 2 targets fine-grained position-specific citation prediction. To enhance these two tasks, we build a dataset containing 7,267 instances for Task 1 and 8,541 instances for Task 2, enabling comprehensive evaluation of both retrieval and generation. (2) We construct a three-level large-scale corpus with 554k papers spanning many major subfields, using an incremental pipeline. (3) We propose a multi-level hybrid RAG approach to citation prediction, fine-tuning embedding models with contrastive learning to capture complex citation relationships, paired with specialized generation models. (4) We conduct extensive experiments across state-of-the-art language models, including closed-source APIs, open-source models, and our fine-tuned generators, demonstrating the effectiveness of our framework. Our open-source toolkit enables reproducible evaluation and focuses on academic literature, providing the first comprehensive evaluation framework for citation prediction and serving as a methodological template for other scientific domains. Our source code and data are released at https://github.com/LQgdwind/CiteRAG.

Manipulating transparent objects presents significant challenges due to the complexities introduced by their reflection and refraction properties, which considerably hinder the accurate estimation of their 3D shapes. To address these challenges, we propose a single-view RGB-D-based depth completion framework, TransDiff, that leverages the Denoising Diffusion Probabilistic Models(DDPM) to achieve material-agnostic object grasping in desktop. Specifically, we leverage features extracted from RGB images, including semantic segmentation, edge maps, and normal maps, to condition the depth map generation process. Our method learns an iterative denoising process that transforms a random depth distribution into a depth map, guided by initially refined depth information, ensuring more accurate depth estimation in scenarios involving transparent objects. Additionally, we propose a novel training method to better align the noisy depth and RGB image features, which are used as conditions to refine depth estimation step by step. Finally, we utilized an improved inference process to accelerate the denoising procedure. Through comprehensive experimental validation, we demonstrate that our method significantly outperforms the baselines in both synthetic and real-world benchmarks with acceptable inference time. The demo of our method can be found on: https://wang-haoxiao.github.io/TransDiff/

A unified foundation model for medical time series—pretrained on open access and ethically reviewed medical corpora—offers the potential to reduce annotation burdens, minimize model customization, and enable robust transfer across clinical institutions, modalities, and tasks, particularly in data-scarce or privacy-constrained environments. However, existing time series foundation models struggle to handle medical time series data due to its inherent challenges, including irregular intervals, heterogeneous sampling rates, and frequent missingness. To address these challenges, we introduce MIRA, a unified foundation model specifically designed for medical time series forecasting. MIRA incorporates a Continuous-Time Rotary Positional Encoding that enables fine-grained modeling of variable time intervals, a frequency-specific mixture-of-experts layer that routes computation across latent frequency regimes to further promote temporal specialization, and a Continuous Dynamics Extrapolation Block based on Neural ODE that models the continuous trajectory of latent states, enabling accurate forecasting at arbitrary target timestamps. Pretrained on a large-scale and diverse medical corpus comprising over 454 billion time points collect from publicly available datasets, MIRA achieving reductions in forecasting errors by an average of 8% and 6% in out-of-distribution and in-distribution scenarios, respectively. We also introduce a comprehensive benchmark spanning multiple downstream clinical tasks, establishing a foundation for future research in medical time series modeling.

Uncrewed aerial vehicle (UAV)-assisted joint communication and localization (JCAL) system have great potential and capacity to make future Internet of Things efficient, safe, smart, reliable, and sustainable. Generally, the traditional UAV path planning methods set the flying duration and hovering duration of UAVs as constants, and ignore the importance of UAV operation time in emergency rescue and other scenarios. In this article, we consider the path planning and time scheduling problem of UAV-assisted JCAL system for minimizing the UAV operation time under the constraints of the localization accuracy, communication message, and energy loss. Specifically, we first formulate path planning and time scheduling problem for UAV-assisted JCAL system and derive Cramér-Rao bound (CRB) as the localization accuracy constraint. The variables in the constraints of localization accuracy, communication overhead, and energy loss are deeply coupled, which leads to nonconvex optimization problems. Next, to solve the high nonconvex problem, we divide the original problem into two subproblems, i.e., time scheduling subproblem and path planning subproblem. We use equivalent convex transformation and successive convex approximation (SCA) to transform the nonconvex constraints into convex forms for solving the subproblems, respectively. Lastly, aiming to the robust problem of target and channel parameters, we convert the robust constraints into convex constraint forms by equivalent proof and S-Procedure. On this basis, we develop a robust algorithm for solving the uncertainty of target and channel parameters. Simulation results verify the feasibility of the proposed methods.

Future vehicle localization will be committed to improving the positioning accuracy and energy efficiency of localization systems in the intelligent transportation. Recently, reconfigurable intelligent surface (RIS) as an emerging technology has gained widespread attention and is favorable to enhance the performance of vehicle localization systems because of its capacity of customizing the wireless channel. In this paper, in order to minimize the transmit power, we consider the joint active and passive beamforming problem of RIS-assisted vehicle localization system under the constraints of the localization accuracy and the phase shift parameters of the RIS. Specifically, we establish the model of RIS-assisted vehicle localization system and derive the Cramér-Rao bound (CRB) as the localization performance metric. Next, for the scenario of single vehicle localization, we derive the optimal RISs’ phases, and obtain the optimal solution for joint active and passive beamforming based on semidefinite programming relaxation of the non-convex beamforming problem and the corresponding equivalent analysis. Lastly, aimming to the scenario of multiple vehicles localization, we transform the nonconvex joint active and passive beamforming problem into semidefinite programming (SDP) and geometric programming (GP) form subproblems through alternating optimization. Simulation results verify the feasibility of the proposed methods.

Multi-target tracking (MTT) plays a significant role in intelligent transportation systems, serving as an enabling technology for applications such as self-driving, surveillance, and navigation. To enhance the MTT performance, the unmanned aerial vehicles (UAVs) have emerged as effective assistants to MIMO radar system, due to their advantages of high flexibility, controllable deployment and cost-effectiveness. Towards this end, this work investigates a multi-UAV-enabled MIMO radar system, in which each UAV is equipped with a MIMO radar unit and dispatched to track multiple targets simultaneously. We are interested in the joint trajectory planning and transmit resource optimization (i.e. radar waveform optimization and transmit power allocation) to minimize the system power consumption, subject to constraints related to UAVs motion, system resources, and tracking accuracy. Specifically, the posterior Cramér-Rao Lower Bound (PCRLB) is derived and employed as a guideline for the joint optimization. Given the non-convex and inter-variable coupling nature of the formulated problem, we decompose it into three sub-problems and design an alternating optimization method. Firstly, for the UAVs trajectory planning, we obtain sub-optimal results leveraging the successive convex approximation (SCA)-based algorithm. Next, we present a feasible solution set for radar waveform optimization. For transmit power allocation, we perform a convex transformation and find the numerical solution. In addition, through introducing the Lagrange dual method, we further obtain the optimal analytical solution. Finally, simulation results demonstrate the effectiveness and advantages of the developed strategy.

In recent years, graph neural networks (GNN) have developed rapidly in various fields, but the high computational consumption of its model training often discourages some graph owners who want to train GNN models but lack computing power. Therefore, these data owners often cooperate with external calculators during the model training process, which will raise critical severe privacy concerns. Protecting private information in graph, however, is difficult due to the complex graph structure consisting of node features and edges. To solve this problem, we propose a new privacy-preserving GNN named MDP based on matrix decomposition and differential privacy (DP), which allows external calculators train GNN models without knowing the original data. Specifically, we first introduce the concept of topological secret sharing (TSS), and design a novel matrix decomposition method named eigenvalue selection (ES) according to TSS, which can preserve the message passing ability of adjacency matrix while hiding edge information. We evaluate the feasibility and performance of our model through extensive experiments, which demonstrates that MDP model achieves accuracy comparable to the original model, with practically affordable overhead.

Recently, knowledge-grounded conversations in the open domain gain great attention from researchers. Existing works on retrieval-based dialogue systems have paid tremendous efforts to utilize neural networks to build a matching model, where all of the context and knowledge contents are used to match the response candidate with various representation methods.