arXiv subjects

Xiaoqian Wang

Publications and source records attributed to Xiaoqian Wang.

At least 19 recordsLinked to original sources

arXiv2026

Uncertainty-Aware Hybrid Retrieval for Long-Document RAG

Retrieval augmented generation (RAG) depends critically on the quality and granularity of retrieved evidence. Large retrieval units preserve context but often introduce irrelevant content, which can dilute answer bearing evidence and worsen long context utilization. Fine-grained units are more compact, but they may be difficult to retrieve reliably because short chunks can lack semantic, lexical, or bridging cues needed to match the query. We propose Uncertainty-aware Multi-Granularity RAG (UMG-RAG), a training-free hybrid retrieval framework that treats chunk granularity as query-specific reliability estimation. Instead of training a new retriever or modifying the generator, UMG-RAG uses existing dense and sparse retrievers as complementary experts across multiple chunk granularities. For each query, it converts each expert-granularity score list into an evidence distribution, estimates reliability from distribution entropy, and fuses candidates according to query-specific semantic, lexical, and granularity confidence. We further introduce UMGP-RAG, a parent promotion variant that uses fine-grained hits to locate relevant evidence while returning broader non-redundant parent chunks for local coherence. Experiments on question answering benchmarks show that uncertainty-aware fusion and parent promotion improve generation quality while maintaining a lightweight, plug-and-play retrieval pipeline.

cs.AI↗

arXiv2026

When the Same Coefficients Reach Different Places: Asymmetric Realizability in Transplanting Tokenizers across Large Language Models

Tokenizer transplant in cross-vocabulary model composition reconstructs donor-only embedding rows as weighted combinations over shared lexical anchors and reuses those coefficients on the base. We identify a structural geometric property of this reconstruction: the same coefficient vector reaches different sets in the donor and base anchor spans, an \emph{asymmetric realizability} gap. Across 65 donor-base pairs under OMP, with cross-operator validation on CLP, WECHSEL, and FOCUS, we construct \textit{breaker tokens}: single coefficient vectors that remain statistically inert in the donor anchor span while producing a high-salience reconstruction in the base. The same Gemma-2-2B donor checkpoint admits this construction against 13 different downstream bases drawn from five model families. The planted direction passes weight-merging with a clean reference unchanged. In a deployer case study, standard LoRA fine-tuning suppresses the breaker primarily on prompts whose distribution matches the training corpus and is not a sufficient mitigation against this attack family in our setting. The tested spectral filters miss the asymmetry. We discuss potential misuse in the open-weight composition supply chain.

cs.LG↗

arXiv2026

The Vision Wormhole: Latent-Space Communication in Heterogeneous Multi-Agent Systems

Multi-Agent Systems (MAS) powered by Large Language Models have unlocked advanced collaborative reasoning, yet they remain bottlenecked by discrete text communication, which imposes runtime overhead and information quantization loss. While latent state transfer offers an alternative, existing approaches either assume homogeneous sender--receiver architectures or rely on pair-specific learned translators, limiting scalability across diverse model families with disjoint manifolds. We reconceptualize the visual interface of Vision-Language Models (VLMs), trained for natural images, as a continuous communication channel between heterogeneous agents, and instantiate this idea as the \textbf{Vision Wormhole}: a Universal Visual Codec maps reasoning traces into a shared continuous reference space and injects them into the receiver's visual pathway, yielding cross-architecture latent state transfer without per-pair translators. The framework adopts a hub-and-spoke topology that reduces alignment complexity from $O(N^2)$ to $O(N)$, and is trained by label-free teacher--student distillation against the text channel, requiring no parallel hidden-state supervision. Extensive experiments across heterogeneous VLM families (Qwen-VL, Gemma, SmolVLM2, LFM2.5-VL) and nine reasoning benchmarks show that the Vision Wormhole reduces end-to-end wall-clock time across most evaluated settings and yields positive macro-average $Δ$-accuracy.

cs.CL↗

arXiv2026

Do Proactive Agents Really Need an LLM to Decide When to Wake and What to Anchor?

Proactive agents read user activity as text and call an LLM on every event to decide whether to act. But user activity is not natively text: it is a structured event stream of (actor, verb, object, timestamp) tuples that the operating system already maintains in graph form. Rendering the structure as text and asking an LLM to recover it is a round-trip the system never had to take. We treat the always-on signal as graph updates rather than text and use a small temporal-graph-learning (TGL) model as the encoder: one forward pass yields a per-event trigger probability and a per-entity routing score, and only the downstream agent (turning a small structured handoff into a fluent user-facing sentence) is an LLM call, invoked only when the trigger fires. TGL improves F1 on each of 14 backbones (mean +16.7, up to +46.0); in trigger-architecture comparisons, one TGL checkpoint gives the strongest trigger AUCs and the most stable deployed threshold. It runs at 11.13 ms per event on a GPU server and 13.99 ms on a consumer laptop, approximately 4--7x and 12--83x faster than every single-forward LLM-as-trigger configuration tested in each regime, with an approximately 220 MiB BF16 resident footprint deployable on-device alongside the privacy-sensitive activity stream it consumes.

cs.CL↗

arXiv2026

The Cost of Context: Mitigating Textual Bias in Multimodal Retrieval-Augmented Generation

While Multimodal Large Language Models (MLLMs) are increasingly integrated with Retrieval-Augmented Generation (RAG) to mitigate hallucinations, the introduction of external documents can conceal severe failure modes at the instance level. We identify and formalize the phenomenon of recorruption, where the introduction of even perfectly accurate "oracle" context causes a capable model to abandon an initially correct prediction. Through a mechanistic diagnosis of internal attention matrices, we show that recorruption is driven by a two-fold attentional collapse: (1) visual blindness, characterized by the systemic suppression of visual attention mass ($M_{vis}$) and sharpness ($S_{vis}$), and (2) a structural positional bias that forces the model to prioritize boundary tokens over semantic relevance. Our analysis reveals an Illusion of Success, demonstrating that many seemingly correct RAG outcomes are merely positional coincidences where the model's textual copying bias happens to align with the ground-truth location. To address these vulnerabilities, we propose Bottleneck Attention Intervention for Recovery (BAIR), a parameter-free, inference-time framework that restores visual saliency and applies position-aware penalties to textual distractors. Across medical factuality, social fairness, and geospatial benchmarks, BAIR successfully restores multimodal grounding and improves diagnostic reliability without requiring model retraining or fine-tuning.

cs.CL↗

arXiv2026

Online conformal inference for multi-step time series forecasting

We consider the problem of constructing distribution-free prediction intervals for multi-step time series forecasting, with a focus on the temporal dependencies inherent in multi-step forecast errors. We establish that the optimal $h$-step-ahead forecast errors exhibit serial correlation up to lag $(h-1)$ under a general non-stationary autoregressive data generating process. To leverage these properties, we propose the Autocorrelated Multi-step Conformal Prediction (AcMCP) method, which effectively incorporates autocorrelations in multi-step forecast errors, resulting in more statistically efficient prediction intervals. This method guarantees asymptotic marginal coverage for multi-step prediction intervals, though we note that, for finite samples, the coverage error admits an upper bound that increases with the forecasting horizon. Additionally, we extend several easy-to-implement conformal prediction methods, originally designed for single-step forecasting, to accommodate multi-step scenarios. Through empirical evaluations, including simulations and applications to data, we demonstrate that AcMCP achieves coverage that closely aligns with the target within local windows, while providing adaptive prediction intervals that effectively respond to varying conditions.

stat.ME↗

arXiv2026

On the Effects of Adversarial Perturbations on Distribution Robustness

Adversarial robustness refers to a model's ability to resist perturbation of inputs, while distribution robustness evaluates the performance of the model under data shifts. Although both aim to ensure reliable performance, prior work has revealed a tradeoff in distribution and adversarial robustness. Specifically, adversarial training might increase reliance on spurious features, which can harm distribution robustness, especially the performance on some underrepresented subgroups. We present a theoretical analysis of adversarial and distribution robustness that provides a tractable surrogate for per-step adversarial training by studying models trained on perturbed data. In addition to the tradeoff, our work further identified a nuanced phenomenon that $\ell_\infty$ perturbations on data with moderate bias can yield an increase in distribution robustness. Moreover, the gain in distribution robustness remains on highly skewed data when simplicity bias induces reliance on the core feature, characterized as greater feature separability. Our theoretical analysis extends the understanding of the tradeoff by highlighting the interplay of the tradeoff and the feature separability. Despite the tradeoff that persists in many cases, overlooking the role of feature separability may lead to misleading conclusions about robustness.

cs.LG↗

arXiv2025

Distributed Online Economic Dispatch with Time-Varying Coupled Inequality Constraints

We investigate the distributed online economic dispatch problem for power systems with time-varying coupled inequality constraints. The problem is formulated as a distributed online optimization problem in a multi-agent system. At each time step, each agent only observes its own instantaneous objective function and local inequality constraints; agents make decisions online and cooperate to minimize the sum of the time-varying objectives while satisfying the global coupled constraints. To solve the problem, we propose an algorithm based on the primal-dual approach combined with constraint-tracking. Under appropriate assumptions that the objective and constraint functions are convex, their gradients are uniformly bounded, and the path length of the optimal solution sequence grows sublinearly, we analyze theoretical properties of the proposed algorithm and prove that both the dynamic regret and the constraint violation are sublinear with time horizon T. Finally, we evaluate the proposed algorithm on a time-varying economic dispatch problem in power systems using both synthetic data and Australian Energy Market data. The results demonstrate that the proposed algorithm performs effectively in terms of tracking performance, constraint satisfaction, and adaptation to time-varying disturbances, thereby providing a practical and theoretically well-supported solution for real-time distributed economic dispatch.

math.OC↗

arXiv2025

SUV: Scalable Large Language Model Copyright Compliance with Regularized Selective Unlearning

Large Language Models (LLMs) have transformed natural language processing by learning from massive datasets, yet this rapid progress has also drawn legal scrutiny, as the ability to unintentionally generate copyrighted content has already prompted several prominent lawsuits. In this work, we introduce SUV (Selective Unlearning for Verbatim data), a selective unlearning framework designed to prevent LLM from memorizing copyrighted content while preserving its overall utility. In detail, the proposed method constructs a dataset that captures instances of copyrighted infringement cases by the targeted LLM. With the dataset, we unlearn the content from the LLM by means of Direct Preference Optimization (DPO), which replaces the verbatim copyrighted content with plausible and coherent alternatives. Since DPO may hinder the LLM's performance in other unrelated tasks, we integrate gradient projection and Fisher information regularization to mitigate the degradation. We validate our approach using a large-scale dataset of 500 famous books (predominantly copyrighted works) and demonstrate that SUV significantly reduces verbatim memorization with negligible impact on the performance on unrelated tasks. Extensive experiments on both our dataset and public benchmarks confirm the scalability and efficacy of our approach, offering a promising solution for mitigating copyright risks in real-world LLM applications.

cs.CL↗

arXiv2025

Inland Waterway Object Detection in Multi-environment: Dataset and Approach

The success of deep learning in intelligent ship visual perception relies heavily on rich image data. However, dedicated datasets for inland waterway vessels remain scarce, limiting the adaptability of visual perception systems in complex environments. Inland waterways, characterized by narrow channels, variable weather, and urban interference, pose significant challenges to object detection systems based on existing datasets. To address these issues, this paper introduces the Multi-environment Inland Waterway Vessel Dataset (MEIWVD), comprising 32,478 high-quality images from diverse scenarios, including sunny, rainy, foggy, and artificial lighting conditions. MEIWVD covers common vessel types in the Yangtze River Basin, emphasizing diversity, sample independence, environmental complexity, and multi-scale characteristics, making it a robust benchmark for vessel detection. Leveraging MEIWVD, this paper proposes a scene-guided image enhancement module to improve water surface images based on environmental conditions adaptively. Additionally, a parameter-limited dilated convolution enhances the representation of vessel features, while a multi-scale dilated residual fusion method integrates multi-scale features for better detection. Experiments show that MEIWVD provides a more rigorous benchmark for object detection algorithms, and the proposed methods significantly improve detector performance, especially in complex multi-environment scenarios.

cs.CV↗

arXiv2024

A Unified Debiasing Approach for Vision-Language Models across Modalities and Tasks

Recent advancements in Vision-Language Models (VLMs) have enabled complex multimodal tasks by processing text and image data simultaneously, significantly enhancing the field of artificial intelligence. However, these models often exhibit biases that can skew outputs towards societal stereotypes, thus necessitating debiasing strategies. Existing debiasing methods focus narrowly on specific modalities or tasks, and require extensive retraining. To address these limitations, this paper introduces Selective Feature Imputation for Debiasing (SFID), a novel methodology that integrates feature pruning and low confidence imputation (LCI) to effectively reduce biases in VLMs. SFID is versatile, maintaining the semantic integrity of outputs and costly effective by eliminating the need for retraining. Our experimental results demonstrate SFID's effectiveness across various VLMs tasks including zero-shot classification, text-to-image retrieval, image captioning, and text-to-image generation, by significantly reducing gender biases without compromising performance. This approach not only enhances the fairness of VLMs applications but also preserves their efficiency and utility across diverse scenarios.

cs.CV↗

arXiv2024

A liquid crystal geometric phase hyperbolic lens with a positive focal length for arbitrary circularly polarized incidence

This paper presents the design and experimental validation of a liquid crystal geometric phase hyperbolic lens (LCHL) with positive focal lengths for arbitrary circularly polarized light. Utilizing Pancharatnam-Berry phase modulation, the lens enables isotropic focusing for both left- and right-handed circular polarization. We fabricated and characterized two lenses with different focal lengths, and their diffraction patterns were analyzed experimentally. The results exhibit strong agreement with theoretical simulations, highlighting the lens's capability for precise optical field modulation. The proposed LCHL demonstrates significant potential for detecting weak polarization states, making it a promising tool for advanced applications in biomedical imaging, optical vortex generation, and multifocal optical systems.

physics.optics↗

arXiv2024

Alpha-Proton Differential Flow of A Coronal Mass Ejection at 15 Solar Radii

Alpha-proton differential flow ($V_{αp}$) of coronal mass ejections (CMEs) and solar wind from the Sun to 1 au and beyond could influence the instantaneous correspondence of absolute abundances of alpha particles (He$^{2+}$/H$^{+}$) between solar corona and interplanetary space as the abundance of a coronal source can vary with time. Previous studies based on Ulysses and Helios showed that $V_{αp}$ is negligible within CMEs from 5 to 0.3 au, similar to slow solar wind ($<$ 400 km s$^{-1}$). However, recent new observations using Parker Solar Probe (PSP) revealed that the $V_{αp}$ of slow wind increases to $\sim$60 km s$^{-1}$ inside 0.1 au. It is significant to answer whether the $V_{αp}$ of CMEs exhibits the similar behavior near the Sun. In this Letter, we report the $V_{αp}$ of a CME measured by PSP at $\sim$15 $R_\odot$ for the first time, which demonstrates that the $V_{αp}$ of CMEs is obvious and complex inside 0.1 au while keeps lower than the local Alfvén speed. A very interesting point is that the same one CME duration can be divided into A and B intervals clearly with Coulomb number below and beyond 0.5, respectively. The means of $V_{αp}$ and alpha-to-proton temperature ratios of interval A (B) is 96.52 (21.96) km s$^{-1}$ and 7.65 (2.23), respectively. This directly illustrates that Coulomb collisions play an important role in reducing the non-equilibrium features of CMEs. Our study indicates that the absolute elemental abundances of CMEs also might vary during their propagation.

astro-ph.SR↗

arXiv2024

Learning the irreversible progression trajectory of Alzheimer's disease

Alzheimer's disease (AD) is a progressive and irreversible brain disorder that unfolds over the course of 30 years. Therefore, it is critical to capture the disease progression in an early stage such that intervention can be applied before the onset of symptoms. Machine learning (ML) models have been shown effective in predicting the onset of AD. Yet for subjects with follow-up visits, existing techniques for AD classification only aim for accurate group assignment, where the monotonically increasing risk across follow-up visits is usually ignored. Resulted fluctuating risk scores across visits violate the irreversibility of AD, hampering the trustworthiness of models and also providing little value to understanding the disease progression. To address this issue, we propose a novel regularization approach to predict AD longitudinally. Our technique aims to maintain the expected monotonicity of increasing disease risk during progression while preserving expressiveness. Specifically, we introduce a monotonicity constraint that encourages the model to predict disease risk in a consistent and ordered manner across follow-up visits. We evaluate our method using the longitudinal structural MRI and amyloid-PET imaging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Our model outperforms existing techniques in capturing the progressiveness of disease risk, and at the same time preserves prediction accuracy.

cs.LG↗

arXiv2024

SHIELD: Evaluation and Defense Strategies for Copyright Compliance in LLM Text Generation

Large Language Models (LLMs) have transformed machine learning but raised significant legal concerns due to their potential to produce text that infringes on copyrights, resulting in several high-profile lawsuits. The legal landscape is struggling to keep pace with these rapid advancements, with ongoing debates about whether generated text might plagiarize copyrighted materials. Current LLMs may infringe on copyrights or overly restrict non-copyrighted texts, leading to these challenges: (i) the need for a comprehensive evaluation benchmark to assess copyright compliance from multiple aspects; (ii) evaluating robustness against safeguard bypassing attacks; and (iii) developing effective defense targeted against the generation of copyrighted text. To tackle these challenges, we introduce a curated dataset to evaluate methods, test attack strategies, and propose lightweight, real-time defense to prevent the generation of copyrighted text, ensuring the safe and lawful use of LLMs. Our experiments demonstrate that current LLMs frequently output copyrighted text, and that jailbreaking attacks can significantly increase the volume of copyrighted output. Our proposed defense mechanism significantly reduces the volume of copyrighted text generated by LLMs by effectively refusing malicious requests. Code is publicly available at https://github.com/xz-liu/SHIELD

cs.CL↗

arXiv2024

Another look at forecast trimming for combinations: robustness, accuracy and diversity

Forecast combination is widely recognized as a preferred strategy over forecast selection due to its ability to mitigate the uncertainty associated with identifying a single "best" forecast. Nonetheless, sophisticated combinations are often empirically dominated by simple averaging, which is commonly attributed to the weight estimation error. The issue becomes more problematic when dealing with a forecast pool containing a large number of individual forecasts. In this paper, we propose a new forecast trimming algorithm to identify an optimal subset from the original forecast pool for forecast combination tasks. In contrast to existing approaches, our proposed algorithm simultaneously takes into account the robustness, accuracy and diversity issues of the forecast pool, rather than isolating each one of these issues. We also develop five forecast trimming algorithms as benchmarks, including one trimming-free algorithm and several trimming algorithms that isolate each one of the three key issues. Experimental results show that our algorithm achieves superior forecasting performance in general in terms of both point forecasts and prediction intervals. Nevertheless, we argue that diversity does not always have to be addressed in forecast trimming. Based on the results, we offer some practical guidelines on the selection of forecast trimming algorithms for a target series.

stat.ME↗

arXiv2024

Comparison of Ion-Proton Differential Speed between ICMEs and Solar Wind near 1 au

The elemental abundance of ICMEs and solar wind near 1 au is often adopted to represent the abundance in the corresponding coronal sources. However, the absolute abundance of heavy ions (relative to hydrogen) near 1 au might be different from the coronal abundance due to the ion-proton differential speed ($V_{ip}$). To illustrate the $V_{ip}$ characteristics and explore whether it influences the absolute abundance analysis for ICMEs and solar wind, we perform a statistical study on the $V_{ip}$ for He$^{2+}$, C$^{5+}$, O$^{6+}$, and Fe$^{10+}$ in both ICMEs and solar wind based on measurements of Advanced Composition Explorer. The results show that the $V_{ip}$ is negligible within ICMEs and slow solar wind ($<$ 400 km s$^{-1}$), while obvious in the intermediate (400 -- 600 km s$^{-1}$) and fast wind ($>$ 600 km s$^{-1}$). Previous studies showed that the $V_{ip}$ in ICMEs keeps negligible during propagation from 0.3 to 5 au, but in solar wind it increases with the decreasing heliocentric distance. Therefore, it might be questionable to infer the absolute abundance of coronal sources through in-situ abundance near 1 au for solar wind. Fortunately, the ion-oxygen (O$^{6+}$) differential speed ($V_{io}$) is negligible for He$^{2+}$, C$^{5+}$, and Fe$^{10+}$ within both ICMEs and solar wind, and previous studies suggested that the $V_{io}$ does not vary significantly with the heliocentric distance. This indicates that various heavy ions always flow at the same bulk speed and their relative abundance (relative to oxygen) near 1 au can represent the coronal abundance for both ICMEs and solar wind.

astro-ph.SR↗

arXiv2024

Distributed ARIMA Models for Ultra-long Time Series

Providing forecasts for ultra-long time series plays a vital role in various activities, such as investment decisions, industrial production arrangements, and farm management. This paper develops a novel distributed forecasting framework to tackle challenges associated with forecasting ultra-long time series by using the industry-standard MapReduce framework. The proposed model combination approach facilitates distributed time series forecasting by combining the local estimators of time series models delivered from worker nodes and minimizing a global loss function. In this way, instead of unrealistically assuming the data generating process (DGP) of an ultra-long time series stays invariant, we make assumptions only on the DGP of subseries spanning shorter time periods. We investigate the performance of the proposed approach with AutoRegressive Integrated Moving Average (ARIMA) models using the real data application as well as numerical simulations. Compared to directly fitting the whole data with ARIMA models, our approach results in improved forecasting accuracy and computational efficiency both in point forecasts and prediction intervals, especially for longer forecast horizons. Moreover, we explore some potential factors that may affect the forecasting performance of our approach.

stat.AP↗