arXiv subjects

Ruiqi Wang

Publications and source records attributed to Ruiqi Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

On the Effectiveness of Context Compression for Repository-Level Tasks: An Empirical Investigation

Repository-level code intelligence tasks require large language models (LLMs) to process long, multi-file contexts. Such inputs introduce three challenges: crucial context can be obscured by noise, truncated due to limited windows, and increased inference latency. Context compression mitigates these risks by condensing inputs. While studied in NLP, its applicability to code tasks remains largely unexplored. We present the first systematic empirical study of context compression for repository-level code intelligence, organizing eight methods into three paradigms: discrete token sequences, continuous latent vectors, and visual tokens. We evaluate them on code completion and generation, measuring performance and efficiency. Results show context compression is effective: at 4x compression, continuous latent vector methods surpass full-context performance by up to 28.3% in BLEU score, indicating they filter noise rather than just truncating. On efficiency, all paradigms reduce inference cost. Both visual and text-based compression achieve up to 50% reduction in end-to-end latency at high ratios, approaching the cost of inference without repository context. These findings establish context compression as a viable approach and provide guidance for paradigm selection.

cs.SE↗

arXiv2026

Pre-Execution Safety Gate & Task Safety Contracts for LLM-Controlled Robot Systems

Large Language Models (LLMs) are increasingly used to convert task commands into robot-executable code, however this pipeline lacks validation gates to detect unsafe and defective commands before they are translated into robot code. Furthermore, even commands that appear safe at the outset can produce unsafe state transitions during execution in the absence of continuous constraint monitoring. In this research, we introduce SafeGate, a neurosymbolic safety architecture that prevents unsafe natural language task commands from reaching robot execution. Drawing from ISO 13482 safety standard, SafeGate extracts structured safety-relevant properties from natural language commands and applies a deterministic decision gate to authorize or reject execution. In addition, we introduce Task Safety Contracts, which decomposes commands that pass through the gate into invariants, guards, and abort conditions to prevent unsafe state transitions during execution. We further incorporate Z3 SMT solving to enforce constraint checking derived from the Task Safety Contracts. We evaluate SafeGate against existing LLM-based robot safety frameworks and baseline LLMs across 230 benchmark tasks, 30 AI2-THOR simulation scenarios, and real-world robot experiments. Results show that SafeGate significantly reduces the acceptance of defective commands while maintaining a high acceptance of benign tasks, demonstrating the importance of pre-execution safety gates for LLM-controlled robot systems

cs.RO↗

arXiv2026

Millimeter-Wave RIS: Hardware Design and System-Level Considerations

Reconfigurable intelligent surfaces have emerged as a promising hardware platform for shaping wireless propagation environments at millimeter-wave (mm-Wave) frequencies and beyond. While many existing studies emphasize channel modeling and signal processing, practical RIS deployment is fundamentally governed by hardware design choices and their system-level implications. This paper presents a hardware-centric overview of recent mm-Wave RIS developments, covering wideband realizations, high-resolution phase-quantized designs, fully printed low-cost implementations, optically transparent surfaces, RIS-on-chip solutions, and emerging three-dimensional architectures. Key challenges including mutual coupling, calibration, multi-RIS interaction, and frequency-dependent phase control are discussed to bridge hardware realization with system-level optimization. This overview provides practical design insights and aims to guide future RIS research toward scalable, efficient, and practically deployable intelligent surface architectures.

eess.SY↗

arXiv2026

AlignSentinel: Alignment-Aware Detection of Prompt Injection Attacks

Prompt injection attacks insert malicious instructions into an LLM's input to steer it toward an attacker-chosen task instead of the intended one. Existing detection defenses typically classify any input with instruction as malicious, leading to misclassification of benign inputs containing instructions that align with the intended task. In this work, we account for the instruction hierarchy and distinguish among three categories: inputs with misaligned instructions, inputs with aligned instructions, and non-instruction inputs. We introduce AlignSentinel, a three-class classifier that leverages features derived from LLM's attention maps to categorize inputs accordingly. To support evaluation, we construct the first systematic benchmark containing inputs from all three categories. Experiments on both our benchmark and existing ones--where inputs with aligned instructions are largely absent--show that AlignSentinel accurately detects inputs with misaligned instructions and substantially outperforms baselines.

cs.CR↗

arXiv2026

3-D Reconfigurable Intelligent Surface: From Reflection to Transmission and From Single Hemisphere to Full 3-D Coverage

Reconfigurable intelligent surfaces (RIS) are conventionally implemented as two-dimensional (2D) electromagnetic (EM) structures to steer incident waves toward desired reflection angles. This approach limits the reflection to a single hemisphere, and the beam-scanning range is relatively small. In this work, a novel three-dimensional (3D) RIS concept is proposed, where beam-scanning can be realized not only through reflection from the illuminated surface but also through controlled transmission toward adjacent surfaces, enabling near blind-spot-free coverage in the full 3D spatial domain. A cube-based 3D-RIS design operating at millimeter-wave (mm-Wave) frequencies and consisting of six interconnected RIS surfaces is presented. Each surface integrates reconfigurable receiving and reflecting arrays with orthogonal polarizations to ensure intrinsic EM isolation, while a reconfigurable feeding network supports dynamic operation. A subarray-based synthesis approach with binary amplitude gating and predefined phase offsets is developed through a unified theoretical model. This model, validated through full-wave simulations, enables efficient beam switching through a shared aperture. Based on this framework, an 8 x 12 element surface comprising six 4 x 4 subarrays is designed, with each surface covering an angular range from -30 deg to +30 deg. The experimental prototype has been characterized in the 24 to 30 GHz band, and the results demonstrate a gain enhancement of 14.7 dB for reflection, while 14.1 dB is achieved for transmission to the neighboring surface. Finally, wireless communication trials using the Pluto software-defined radio platform combined with frequency up/down converters confirm improved constellation quality and a 6-7 dB improvement in error vector magnitude (EVM) for both reflection and neighboring surface transmission scenarios.

eess.SY↗

arXiv2026

MMTS-BENCH: A Comprehensive Benchmark for Time Series Understanding and Reasoning

Time series data are central to domains such as finance, healthcare, and cloud computing, yet existing benchmarks for evaluating various large language models (LLMs) on temporal tasks remain scattered and unsystematic. To bridge this gap, we introduce MMTS-BENCH, a comprehensive multimodal benchmark built upon a hierarchical taxonomy of time-series tasks, spanning structural awareness, feature analysis, temporal reasoning, sequence matching and cross-modal alignment. MMTS-BENCH comprises 2,424 time series question answering (TSQA) pairs across 4 subsets: Base, InWild, Match, and Align, generated through a progressive real-world QA framework and modular synthetic data construction. We conduct extensive evaluations on closed-source, open-source LLMs and existing time series adapted large language models (TS-LLMs), revealing that: (1) TS-LLMs significantly lag behind general-purpose LLMs in cross-domain generalization, (2) LLMs show weaknesses in local tasks compared to global tasks, (3) chain-of-thought (CoT) reasoning and multimodal integration substantially improve performance, and (4) the dominant factor in existing TS-LLMs remains the backbone network capability rather than the time series encoder design. MMTS-BENCH not only provides a rigorous evaluation framework but also offers clear directions for advancing LLMs toward robust, interpretable, and generalizable time-series reasoning.

cs.DB↗

arXiv2026

CausalTAD: Injecting Causal Knowledge into Large Language Models for Tabular Anomaly Detection

Detecting anomalies in tabular data is critical for many real-world applications, such as credit card fraud detection. With the rapid advancements in large language models (LLMs), state-of-the-art performance in tabular anomaly detection has been achieved by converting tabular data into text and fine-tuning LLMs. However, these methods randomly order columns during conversion, without considering the causal relationships between them, which is crucial for accurately detecting anomalies. In this paper, we present CausalTaD, a method that injects causal knowledge into LLMs for tabular anomaly detection. We first identify the causal relationships between columns and reorder them to align with these causal relationships. This reordering can be modeled as a linear ordering problem. Since each column contributes differently to the causal relationships, we further propose a reweighting strategy to assign different weights to different columns to enhance this effect. Experiments across more than 30 datasets demonstrate that our method consistently outperforms the current state-of-the-art methods. The code for CausalTAD is available at https://github.com/350234/CausalTAD.

cs.LG↗

arXiv2026

Transformer-based Hybrid Beamforming with Dynamic Subarray for Near-Space Airship-Borne Communications

This paper proposes a hybrid beamforming framework for massive multiple-input multiple-output (MIMO) in near-space airship-borne communications. To achieve high energy efficiency (EE) in energy-constraint airships, a dynamic subarray structure is introduced, where each radio frequency chain (RFC) is connected to a disjoint subset of the antennas according to channel state information (CSI). The proposed joint dynamic hybrid beamforming network (DyHBFNet) comprises three key components: 1) An analog beamforming network (ABFNet) that optimizes the analog beamforming matrices and provides auxiliary information for the antenna selection network (ASNet) design, 2) an ASNet that dynamically optimizes the connections between antennas and RFCs, and 3) a digital beamforming network (DBFNet) that optimizes digital beamforming matrices by employing a model-driven weighted minimum mean square error algorithm for improving beamforming performance and convergence speed. The proposed ABFNet, ASNet, and DBFNet are all designed based on advanced Transformer encoders. Simulation results demonstrate that the proposed framework significantly enhances spectral efficiency and EE compared to baseline schemes. Additionally, its robust performance under imperfect CSI makes it a scalable solution for practical implementations.

cs.IT↗

arXiv2026

Divergence Results and Convergence of a Variance Reduced Version of ADAM

Stochastic optimization algorithms using exponential moving averages of the past gradients, such as ADAM, RMSProp and AdaGrad, have been having great successes in many applications, especially in training deep neural networks. ADAM in particular stands out as efficient and robust. Despite of its outstanding performance, ADAM has been proved to be divergent for some specific problems. We revisit the divergent question and provide divergent examples under stronger conditions such as in expectation or high probability. Under a variance reduction assumption, we show that an ADAM-type algorithm converges, which means that it is the variance of gradients that causes the divergence of original ADAM. To this end, we propose a variance reduced version of ADAM and provide a convergent analysis of the algorithm. Numerical experiments show that the proposed algorithm has as good performance as ADAM. Our work suggests a new direction for fixing the convergence issues.

cs.LG↗

arXiv2026

A Trainable Optimizer

The concept of learning to optimize involves utilizing a trainable optimization strategy rather than relying on manually defined full gradient estimations such as ADAM. We present a framework that jointly trains the full gradient estimator and the trainable weights of the model. Specifically, we prove that pseudo-linear TO (Trainable Optimizer), a linear approximation of the full gradient, matches SGD's convergence rate while effectively reducing variance. Pseudo-linear TO incurs negligible computational overhead, requiring only minimal additional tensor multiplications. To further improve computational efficiency, we introduce two simplified variants of Pseudo-linear TO. Experiments demonstrate that TO methods converge faster than benchmark algorithms (e.g., ADAM) in both strongly convex and non-convex settings, and fine tuning of an LLM.

cs.LG↗

arXiv2025

Distinguishing Visually Similar Actions: Prompt-Guided Semantic Prototype Modulation for Few-Shot Action Recognition

Few-shot action recognition aims to enable models to quickly learn new action categories from limited labeled samples, addressing the challenge of data scarcity in real-world applications. Current research primarily addresses three core challenges: (1) temporal modeling, where models are prone to interference from irrelevant static background information and struggle to capture the essence of dynamic action features; (2) visual similarity, where categories with subtle visual differences are difficult to distinguish; and (3) the modality gap between visual-textual support prototypes and visual-only queries, which complicates alignment within a shared embedding space. To address these challenges, this paper proposes a CLIP-SPM framework, which includes three components: (1) the Hierarchical Synergistic Motion Refinement (HSMR) module, which aligns deep and shallow motion features to improve temporal modeling by reducing static background interference; (2) the Semantic Prototype Modulation (SPM) strategy, which generates query-relevant text prompts to bridge the modality gap and integrates them with visual features, enhancing the discriminability between similar actions; and (3) the Prototype-Anchor Dual Modulation (PADM) method, which refines support prototypes and aligns query features with a global semantic anchor, improving consistency across support and query samples. Comprehensive experiments across standard benchmarks, including Kinetics, SSv2-Full, SSv2-Small, UCF101, and HMDB51, demonstrate that our CLIP-SPM achieves competitive performance under 1-shot, 3-shot, and 5-shot settings. Extensive ablation studies and visual analyses further validate the effectiveness of each component and its contributions to addressing the core challenges. The source code and models are publicly available at GitHub.

cs.CV↗

arXiv2025

PRIMT: Preference-based Reinforcement Learning with Multimodal Feedback and Trajectory Synthesis from Foundation Models

Preference-based reinforcement learning (PbRL) has emerged as a promising paradigm for teaching robots complex behaviors without reward engineering. However, its effectiveness is often limited by two critical challenges: the reliance on extensive human input and the inherent difficulties in resolving query ambiguity and credit assignment during reward learning. In this paper, we introduce PRIMT, a PbRL framework designed to overcome these challenges by leveraging foundation models (FMs) for multimodal synthetic feedback and trajectory synthesis. Unlike prior approaches that rely on single-modality FM evaluations, PRIMT employs a hierarchical neuro-symbolic fusion strategy, integrating the complementary strengths of large language models and vision-language models in evaluating robot behaviors for more reliable and comprehensive feedback. PRIMT also incorporates foresight trajectory generation, which reduces early-stage query ambiguity by warm-starting the trajectory buffer with bootstrapped samples, and hindsight trajectory augmentation, which enables counterfactual reasoning with a causal auxiliary loss to improve credit assignment. We evaluate PRIMT on 2 locomotion and 6 manipulation tasks on various benchmarks, demonstrating superior performance over FM-based and scripted baselines.

cs.RO↗

arXiv2025

LiNeXt: Revisiting LiDAR Completion with Efficient Non-Diffusion Architectures

3D LiDAR scene completion from point clouds is a fundamental component of perception systems in autonomous vehicles. Previous methods have predominantly employed diffusion models for high-fidelity reconstruction. However, their multi-step iterative sampling incurs significant computational overhead, limiting its real-time applicability. To address this, we propose LiNeXt-a lightweight, non-diffusion network optimized for rapid and accurate point cloud completion. Specifically, LiNeXt first applies the Noise-to-Coarse (N2C) Module to denoise the input noisy point cloud in a single pass, thereby obviating the multi-step iterative sampling of diffusion-based methods. The Refine Module then takes the coarse point cloud and its intermediate features from the N2C Module to perform more precise refinement, further enhancing structural completeness. Furthermore, we observe that LiDAR point clouds exhibit a distance-dependent spatial distribution, being densely sampled at proximal ranges and sparsely sampled at distal ranges. Accordingly, we propose the Distance-aware Selected Repeat strategy to generate a more uniformly distributed noisy point cloud. On the SemanticKITTI dataset, LiNeXt achieves a 199.8x speedup in inference, reduces Chamfer Distance by 50.7%, and uses only 6.1% of the parameters compared with LiDiff. These results demonstrate the superior efficiency and effectiveness of LiNeXt for real-time scene completion.

cs.CV↗

arXiv2025

QoSGMAA: A Robust Multi-Order Graph Attention and Adversarial Framework for Sparse QoS Prediction

With the rapid advancement of internet technologies, network services have become critical for delivering diverse and reliable applications to users. However, the exponential growth in the number of available services has resulted in many similar offerings, posing significant challenges in selecting optimal services. Predicting Quality of Service (QoS) accurately thus becomes a fundamental prerequisite for ensuring reliability and user satisfaction. However, existing QoS prediction methods often fail to capture rich contextual information and exhibit poor performance under extreme data sparsity and structural noise. To bridge this gap, we propose a novel architecture, QoSMGAA, specifically designed to enhance prediction accuracy in complex and noisy network service environments. QoSMGAA integrates a multi-order attention mechanism to aggregate extensive contextual data and predict missing QoS values effectively. Additionally, our method incorporates adversarial neural networks to perform autoregressive supervised learning based on transformed interaction matrices. To capture complex, higher-order interactions among users and services, we employ a discrete sampling technique leveraging the Gumbel-Softmax method to generate informative negative samples. Comprehensive experimental validation conducted on large-scale real-world datasets demonstrates that our proposed model significantly outperforms existing baseline methods, highlighting its strong potential for practical deployment in service selection and recommendation scenarios.

cs.LG↗

arXiv2025

Electromagnetically Reconfigurable Fluid Antenna System for Wireless Communications: Design, Modeling, Algorithm, Fabrication, and Experiment

This paper presents the concept, design, channel modeling, beamforming algorithm development, prototype fabrication, and experimental measurement of an electromagnetically reconfigurable fluid antenna system (ER-FAS), in which each FAS array element features electromagnetic (EM) reconfigurability. Unlike most existing FAS works that investigate spatial reconfigurability by adjusting the position and/or orientation of array elements, the proposed ER-FAS enables direct control over the EM characteristics of each element, allowing for dynamic radiation pattern reconfigurability. Specifically, a novel ER-FAS architecture leveraging software-controlled fluidics is proposed, and corresponding wireless channel models are established. Based on this ER-FAS channel model, a low-complexity greedy beamforming algorithm is developed to jointly optimize the analog phase shift and the radiation state of each array element. The accuracy of the ER-FAS channel model and the effectiveness of the beamforming algorithm are validated through (i) full-wave EM simulations and (ii) numerical spectral efficiency evaluations. These results confirm that the proposed ER-FAS significantly enhances spectral efficiency in both near-field and far-field scenarios compared to conventional antenna arrays. To further validate this design, we fabricate prototypes for both the ER-FAS element and array, using Galinstan liquid metal alloy, fluid silver paste, and software-controlled fluidic channels. The simulation results are experimentally validated through prototype measurements conducted in an anechoic chamber. Additionally, several indoor communication experiments using a pair of software-defined radios demonstrate the superior received power and bit error rate performance of the ER-FAS prototype.

eess.SP↗

arXiv2025

Online Learning Based Efficient Resource Allocation for LoRaWAN Network

The deployment of large-scale LoRaWAN networks requires jointly optimizing conflicting metrics like Packet Delivery Ratio (PDR) and Energy Efficiency (EE) by dynamically allocating transmission parameters, including Carrier Frequency, Spreading Factor, and Transmission Power. Existing methods often oversimplify this challenge, focusing on a single metric or lacking the adaptability needed for dynamic channel environments, leading to suboptimal performance. To address this, we propose two online learning-based resource allocation frameworks that intelligently navigate the PDR-EE trade-off. Our foundational proposal, D-LoRa, is a fully distributed framework that models the problem as a Combinatorial Multi-Armed Bandit. By decomposing the joint parameter selection and employing specialized, disaggregated reward functions, D-LoRa dramatically reduces learning complexity and enables nodes to autonomously adapt to network dynamics. To further enhance performance in LoRaWAN networks, we introduce CD-LoRa, a hybrid framework that integrates a lightweight, centralized initialization phase to perform a one-time, quasi-optimal channel assignment and action space pruning, thereby accelerating subsequent distributed learning. Extensive simulations and real-world field experiments demonstrate the superiority of our frameworks, showing that D-LoRa excels in non-stationary environments while CD-LoRa achieves the fastest convergence in stationary conditions. In physical deployments, our methods outperform state-of-the-art baselines, improving PDR by up to 10.8% and EE by 26.1%, confirming their practical effectiveness for scalable and efficient LoRaWAN networks.

cs.NI↗

arXiv2025

A Screen-printed, Silver-nanowire-based Optically Transparent Wideband Reconfigurable Intelligent Surface for 5G mmWave Applications

The reconfigurable intelligent surface (RIS) can help establish secondary line-of-sight links for 5G mmWave and beyond applications to cater for the large path loss and blockage in millimeter-wave direct links. It is desirable that the RISs are optically transparent so that they can be placed on building windows, etc. However, traditional transparent designs with indium tin oxide films have challenges of high loss and low transparency, while metallic meshes suffer from a large feature size for millimeter wave designs. In this letter, we propose an optically transparent RIS using screen-printed silver nanowires (AgNW) and PIN diodes. The AgNW ink has more than 80% optical transparency and a decent conductivity of $5\times10^{6}$ S/m. The RIS array size is 9x20, and each column serially connects 9 unit cells. The RIS can work from 23.5 GHz to 29.5 GHz with 30° angular stability. The measurement results of the RIS show up to 15 dB gain enhancement within 30° angle range, while maintaining high optical transparency and large bandwidth.

physics.optics↗

arXiv2025

An Empirical Study of Knowledge Distillation for Code Understanding Tasks

Pre-trained language models (PLMs) have emerged as powerful tools for code understanding. However, deploying these PLMs in large-scale applications faces practical challenges due to their computational intensity and inference latency. Knowledge distillation (KD), a promising model compression and acceleration technique, addresses these limitations by transferring knowledge from large teacher models to compact student models, enabling efficient inference while preserving most of the teacher models' capabilities. While this technique has shown remarkable success in natural language processing and computer vision domains, its potential for code understanding tasks remains largely underexplored. In this paper, we systematically investigate the effectiveness and usage of KD in code understanding tasks. Our study encompasses two popular types of KD methods, i.e., logit-based and feature-based KD methods, experimenting across eight student models and two teacher PLMs from different domains on three downstream tasks. The experimental results indicate that KD consistently offers notable performance boosts across student models with different sizes compared with standard fine-tuning. Notably, code-specific PLM demonstrates better effectiveness as the teacher model. Among all KD methods, the latest feature-based KD methods exhibit superior performance, enabling student models to retain up to 98% teacher performance with merely 5% parameters. Regarding student architecture, our experiments reveal that similarity with teacher architecture does not necessarily lead to better performance. We further discuss the efficiency and behaviors in the KD process and inference, summarize the implications of findings, and identify promising future directions.

cs.SE↗