arXiv subjects

Zhang

Publications and source records attributed to Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

The democratic detrender: Ensemble-Based Removal of the Nuisance Signal in Stellar Time-Series Photometry

Accurate, precise, and computationally efficient removal of unwanted activity that exists as a combination of periodic, quasi-periodic, and non-periodic systematic trends in time-series photometric data is a critical step in exoplanet transit analysis. Throughout the years, many different modeling methods have been used for this process, often called "detrending." However, there is no community-wide consensus regarding the favored approach. In order to mitigate model dependency, we present an ensemble-based approach to detrending via community-of-models and the democratic detrender: a modular and scalable open-source coding package that implements ensemble detrending. The democratic detrender allows users to select from a number of packaged detrending methods (including cosine filtering, Gaussian processes, and polynomial fits) or provide their own set of detrended light curves via methods of their choosing. It then combines the individually detrended light curves into a single method marginalized light curve. Additionally, the democratic detrender inflates each data point's uncertainty based on the scatter between detrenders, thereby propagating model-selection uncertainty into the final light curve. This ensemble strategy does not guarantee improvement over the single best-performing detrending method, but it substantially reduces the risk of selecting a detrending solution that is poorly calibrated or overfit to noise.

astro-ph.EP↗

arXiv2026

When Is Self-Disclosure Optimal? Incentives and Governance of AI-Generated Content

Generative artificial intelligence (Gen-AI) is reshaping content creation on digital platforms by reducing production costs and enabling scalable output of varying quality. In response, platforms have begun adopting disclosure policies that require creators to label AI-generated content, often supported by imperfect detection and penalties for non-compliance. This paper develops a formal model to study the economic implications of such disclosure regimes. We compare a non-disclosure benchmark, in which the platform alone detects AI usage, with a mandatory self-disclosure regime in which creators strategically choose whether to disclose or conceal AI use under imperfect enforcement. The model incorporates heterogeneous creators, viewer discounting of AI-labeled content, trust penalties following detected non-disclosure, and endogenous enforcement. The analysis shows that disclosure is optimal only when both the value of AI-generated content and its cost-saving advantage are intermediate. As AI capability improves, the platform's optimal enforcement strategy evolves from strict deterrence to partial screening and eventual deregulation. While disclosure reliably increases transparency, it reduces aggregate creator surplus and can suppress high-quality AI content when AI is technologically advanced. Overall, the results characterize disclosure as a strategic governance instrument whose effectiveness depends on technological maturity and trust frictions.

cs.CY↗

arXiv2026

Randomization Boosts KV Caching, Learning Balances Query Load: A Joint Perspective

KV caching is a fundamental technique for accelerating Large Language Model (LLM) inference by reusing key-value (KV) pairs from previous queries, but its effectiveness under limited memory is highly sensitive to the eviction policy. The default Least Recently Used (LRU) eviction algorithm struggles with dynamic online query arrivals, especially in multi-LLM serving scenarios, where balancing query load across workers and maximizing cache hit rate of each worker are inherently conflicting objectives. We give the first unified mathematical model that captures the core trade-offs between KV cache eviction and query routing. Our analysis reveals the theoretical limitations of existing methods and leads to principled algorithms that integrate provably competitive randomized KV cache eviction with learning-based methods to adaptively route queries with evolving patterns, thus balancing query load and cache hit rate. Our theoretical results are validated by extensive experiments across 4 benchmarks and 3 prefix-sharing settings, demonstrating improvements of up to 6.92$\times$ in cache hit rate, 11.96$\times$ reduction in latency, 14.06$\times$ reduction in time-to-first-token (TTFT), and 77.4% increase in throughput over the state-of-the-art methods. Our code is available at https://github.com/fzwark/KVRouting.

cs.LG↗

arXiv2025

Estimating Effects of Long-Term Treatments

Estimating the effects of long-term treatments through A/B testing is challenging. Treatments, such as updates to product functionalities, user interface designs, and recommendation algorithms, are intended to persist within the system for a long duration of time after their initial launches. However, due to the constraints of conducting long-term experiments, practitioners often rely on short-term experimental results to make product launch decisions. It remains open how to accurately estimate the effects of long-term treatments using short-term experimental data. To address this question, we introduce a longitudinal surrogate framework that decomposes the long-term effects into functions based on user attributes, short-term metrics, and treatment assignments. We outline identification assumptions, estimation strategies, inferential techniques, and validation methods under this framework. Empirically, we demonstrate that our approach outperforms existing solutions by using data from two real-world experiments, each involving more than a million users on WeChat, one of the world's largest social networking platforms.

econ.EM↗

arXiv2025

Exploring Protein Language Model Architecture-Induced Biases for Antibody Comprehension

Recent advances in protein language models (PLMs) have demonstrated remarkable capabilities in understanding protein sequences. However, the extent to which different model architectures capture antibody-specific biological properties remains unexplored. In this work, we systematically investigate how architectural choices in PLMs influence their ability to comprehend antibody sequence characteristics and functions. We evaluate three state-of-the-art PLMs-AntiBERTa, BioBERT, and ESM2--against a general-purpose language model (GPT-2) baseline on antibody target specificity prediction tasks. Our results demonstrate that while all PLMs achieve high classification accuracy, they exhibit distinct biases in capturing biological features such as V gene usage, somatic hypermutation patterns, and isotype information. Through attention attribution analysis, we show that antibody-specific models like AntiBERTa naturally learn to focus on complementarity-determining regions (CDRs), while general protein models benefit significantly from explicit CDR-focused training strategies. These findings provide insights into the relationship between model architecture and biological feature extraction, offering valuable guidance for future PLM development in computational antibody design.

cs.LG↗

arXiv2025

Step-Audio-EditX Technical Report

We present Step-Audio-EditX, the first open-source LLM-based audio model excelling at expressive and iterative audio editing encompassing emotion, speaking style, and paralinguistics alongside robust zero-shot text-to-speech (TTS) capabilities. Our core innovation lies in leveraging only large-margin synthetic data, which circumvents the need for embedding-based priors or auxiliary modules. This large-margin learning approach enables both iterative control and high expressivity across voices, and represents a fundamental pivot from the conventional focus on representation-level disentanglement. Evaluation results demonstrate that Step-Audio-EditX surpasses both MiniMax-2.6-hd and Doubao-Seed-TTS-2.0 in emotion editing and other fine-grained control tasks.

cs.CL↗

arXiv2025

The MDW Hα Sky Survey: Data Release 1

The Mittelman-di Cicco-Walker (MDW) H$α$ Sky Survey is an autonomously-operated all-sky narrow-band (3nm) H$α$ imaging survey. The survey was founded by amateur astronomers and the northern sky (Decl. $\geq$ 0$^\circ$) is presented here in its second stage of refinement for academic use. Each 3.6$\times$3.6 sq. deg MDW field has 12 20-minute individual exposures with a pixel scale of 3.6", a typical PSF of 6", and a stack point source depth of 16-17 magnitudes. The northern MDW Survey Data Release 1 (DR1) includes: calibrated and raw mean and individual images, star-removed mean fields, and point source catalogs for all images matched to Data Release 1 of the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS1) and the INT Galactic Plane Survey (IGAPS). Our initial study of H$α$ filament widths finds a typical FWHM of 30-45" in the Lyra region. The matched catalogs (with a median match distance of ~0.5"), combined with our distinctive narrow-band photometry, are used to identify H$α$ variable and excess sources. These initial studies highlight some of the many scientific uses of the MDW H$α$ survey.

astro-ph.IM↗

arXiv2025

Prompts Generalize with Low Data: Non-vacuous Generalization Bounds for Optimizing Prompts with More Informative Priors

Many prompt engineering techniques have been successful in practice, even when optimizing over a large prompt space with with a small amount of task-specific data. Recent work has partially explained this success by showing generalization bounds which apply PAC-Bayes theory to the discrete prompt space, but they are non-vacuous only in data-rich scenarios. We argue that such widespread success can be more fully explained through more carefully considering data- or distribution-dependent perplexity, which acts as an effective prior and steers the optimization towards prompts that are more ``natural'' for the task at hand. We derive novel generalization bounds that are non-vacuous for data-scarce prompt optimization via more useful priors, formally analyzing how perplexity regularization tightens these bounds by limiting exploration. Empirically, we explore both the bounds' effectiveness and the practical benefits of perplexity regularization in improving prompt generalization.

cs.LG↗

arXiv2025

MoE-CE: Enhancing Generalization for Deep Learning based Channel Estimation via a Mixture-of-Experts Framework

Reliable channel estimation (CE) is fundamental for robust communication in dynamic wireless environments, where models must generalize across varying conditions such as signal-to-noise ratios (SNRs), the number of resource blocks (RBs), and channel profiles. Traditional deep learning (DL)-based methods struggle to generalize effectively across such diverse settings, particularly under multitask and zero-shot scenarios. In this work, we propose MoE-CE, a flexible mixture-of-experts (MoE) framework designed to enhance the generalization capability of DL-based CE methods. MoE-CE provides an appropriate inductive bias by leveraging multiple expert subnetworks, each specialized in distinct channel characteristics, and a learned router that dynamically selects the most relevant experts per input. This architecture enhances model capacity and adaptability without a proportional rise in computational cost while being agnostic to the choice of the backbone model and the learning algorithm. Through extensive experiments on synthetic datasets generated under diverse SNRs, RB numbers, and channel profiles, including multitask and zero-shot evaluations, we demonstrate that MoE-CE consistently outperforms conventional DL approaches, achieving significant performance gains while maintaining efficiency.

eess.SP↗

arXiv2025

A Tractable Approach to Massive Communication and Ubiquitous Connectivity in 6G Standardization

The full-scale 6G standardization has attracted considerable recent attention, especially since the first 3GPP-wide 6G workshop held in March 2025. To understand the practical and fundamental values of 6G and facilitate its standardization, it is crucial to explore the theoretical limits of spectrum, energy, and coverage efficiency considering practical hardware and signaling constraints. In this paper, we present a mean-field-approximation-based investigation on two out of six use case scenarios defined by IMT-2030, namely, massive communication and ubiquitous connectivity. Being aware of the limitation in interference cancellation owing to constrained cost and hardware complexity, we investigate the spectrum reuse architecture in both usage scenarios. We propose a tractable spectrum reuse with low signaling overhead consumed for channel estimation and channel state information (CSI) feedback. Our analysis indicates that the massive communication over cellular and device-to-device (D2D) networks can benefit from channel orthogonalization, while it is unnecessary to share the CSI of interfering links. Moreover, deploying relays or movable base stations, e.g. unmanned aerial vehicle, yields substantial energy and spectrum gain for ubiquitous connectivity, despite introducing interference. As such, the mean-field-optimization-based evaluation is expected to positively impact 6G and NextG standardization in 3GPP and other standardization bodies.

eess.SP↗

arXiv2025

Hierarchical Debate-Based Large Language Model (LLM) for Complex Task Planning of 6G Network Management

6G networks have become increasingly complicated due to novel network architecture and newly emerging signal processing and transmission techniques, leading to significant burdens to 6G network management. Large language models (LLMs) have recently been considered a promising technique to equip 6G networks with AI-native intelligence. Different from most existing studies that only consider a single LLM, this work involves a multi-LLM debate-based scheme for 6G network management, where multiple LLMs can collaboratively improve the initial solution sequentially. Considering the complex nature of 6G domain, we propose a novel hierarchical debate scheme: LLMs will first debate the sub-task decomposition, and then debate each subtask step-by-step. Such a hierarchical approach can significantly reduce the overall debate difficulty by sub-task decomposition, aligning well with the complex nature of 6G networks and ensuring the final solution qualities. In addition, to better evaluate the proposed technique, we have defined a novel dataset named 6GPlan, including 110 complex 6G network management tasks and 5000 keyword solutions. Finally, the experiments show that the proposed hierarchical debate can significantly improve performance compared to baseline techniques, e.g. more than 30% coverage rate and global recall rate improvement.

eess.SY↗

arXiv2025

Prompting Wireless Networks: Reinforced In-Context Learning for Power Control

To manage and optimize constantly evolving wireless networks, existing machine learning (ML)- based studies operate as black-box models, leading to increased computational costs during training and a lack of transparency in decision-making, which limits their practical applicability in wireless networks. Motivated by recent advancements in large language model (LLM)-enabled wireless networks, this paper proposes ProWin, a novel framework that leverages reinforced in-context learning to design task-specific demonstration Prompts for Wireless Network optimization, relying on the inference capabilities of LLMs without the need for dedicated model training or finetuning. The task-specific prompts are designed to incorporate natural language descriptions of the task description and formulation, enhancing interpretability and eliminating the need for specialized expertise in network optimization. We further propose a reinforced in-context learning scheme that incorporates a set of advisable examples into task-specific prompts, wherein informative examples capturing historical environment states and decisions are adaptively selected to guide current decision-making. Evaluations on a case study of base station power control showcases that the proposed ProWin outperforms reinforcement learning (RL)-based methods, highlighting the potential for next-generation future wireless network optimization.

eess.SP↗

arXiv2025

Enhancing Large Language Models (LLMs) for Telecommunications using Knowledge Graphs and Retrieval-Augmented Generation

Large language models (LLMs) have made significant progress in general-purpose natural language processing tasks. However, LLMs are still facing challenges when applied to domain-specific areas like telecommunications, which demands specialized expertise and adaptability to evolving standards. This paper presents a novel framework that combines knowledge graph (KG) and retrieval-augmented generation (RAG) techniques to enhance LLM performance in the telecom domain. The framework leverages a KG to capture structured, domain-specific information about network protocols, standards, and other telecom-related entities, comprehensively representing their relationships. By integrating KG with RAG, LLMs can dynamically access and utilize the most relevant and up-to-date knowledge during response generation. This hybrid approach bridges the gap between structured knowledge representation and the generative capabilities of LLMs, significantly enhancing accuracy, adaptability, and domain-specific comprehension. Our results demonstrate the effectiveness of the KG-RAG framework in addressing complex technical queries with precision. The proposed KG-RAG model attained an accuracy of 88% for question answering tasks on a frequently used telecom-specific dataset, compared to 82% for the RAG-only and 48% for the LLM-only approaches.

cs.CL↗

arXiv2025

Understanding 6G through Language Models: A Case Study on LLM-aided Structured Entity Extraction in Telecom Domain

Knowledge understanding is a foundational part of envisioned 6G networks to advance network intelligence and AI-native network architectures. In this paradigm, information extraction plays a pivotal role in transforming fragmented telecom knowledge into well-structured formats, empowering diverse AI models to better understand network terminologies. This work proposes a novel language model-based information extraction technique, aiming to extract structured entities from the telecom context. The proposed telecom structured entity extraction (TeleSEE) technique applies a token-efficient representation method to predict entity types and attribute keys, aiming to save the number of output tokens and improve prediction accuracy. Meanwhile, TeleSEE involves a hierarchical parallel decoding method, improving the standard encoder-decoder architecture by integrating additional prompting and decoding strategies into entity extraction tasks. In addition, to better evaluate the performance of the proposed technique in the telecom domain, we further designed a dataset named 6GTech, including 2390 sentences and 23747 words from more than 100 6G-related technical publications. Finally, the experiment shows that the proposed TeleSEE method achieves higher accuracy than other baseline techniques, and also presents 5 to 9 times higher sample processing speed.

cs.CL↗

arXiv2025

The Geography of Transportation Cybersecurity: Visitor Flows, Industry Clusters, and Spatial Dynamics

The rapid evolution of the transportation cybersecurity ecosystem, encompassing cybersecurity, automotive, and transportation and logistics sectors, will lead to the formation of distinct spatial clusters and visitor flow patterns across the US. This study examines the spatiotemporal dynamics of visitor flows, analyzing how socioeconomic factors shape industry clustering and workforce distribution within these evolving sectors. To model and predict visitor flow patterns, we develop a BiTransGCN framework, integrating an attention-based Transformer architecture with a Graph Convolutional Network backbone. By integrating AI-enabled forecasting techniques with spatial analysis, this study improves our ability to track, interpret, and anticipate changes in industry clustering and mobility trends, thereby supporting strategic planning for a secure and resilient transportation network. It offers a data-driven foundation for economic planning, workforce development, and targeted investments in the transportation cybersecurity ecosystem.

cs.CY↗

arXiv2025

Do Language Model Agents Align with Humans in Rating Visualizations? An Empirical Study

Large language models encode knowledge in various domains and demonstrate the ability to understand visualizations. They may also capture visualization design knowledge and potentially help reduce the cost of formative studies. However, it remains a question whether large language models are capable of predicting human feedback on visualizations. To investigate this question, we conducted three studies to examine whether large model-based agents can simulate human ratings in visualization tasks. The first study, replicating a published study involving human subjects, shows agents are promising in conducting human-like reasoning and rating, and its result guides the subsequent experimental design. The second study repeated six human-subject studies reported in literature on subjective ratings, but replacing human participants with agents. Consulting with five human experts, this study demonstrates that the alignment of agent ratings with human ratings positively correlates with the confidence levels of the experts before the experiments. The third study tests commonly used techniques for enhancing agents, including preprocessing visual and textual inputs, and knowledge injection. The results reveal the issues of these techniques in robustness and potential induction of biases. The three studies indicate that language model-based agents can potentially simulate human ratings in visualization experiments, provided that they are guided by high-confidence hypotheses from expert evaluators. Additionally, we demonstrate the usage scenario of swiftly evaluating prototypes with agents. We discuss insights and future directions for evaluating and improving the alignment of agent ratings with human ratings. We note that simulation may only serve as complements and cannot replace user studies.

cs.HC↗

arXiv2025

Variability-Driven User-Story Generation using LLM and Triadic Concept Analysis

A widely used Agile practice for requirements is to produce a set of user stories (also called ``agile product backlog''), which roughly includes a list of pairs (role, feature), where the role handles the feature for a certain purpose. In the context of Software Product Lines, the requirements for a family of similar systems is thus a family of user-story sets, one per system, leading to a 3-dimensional dataset composed of sets of triples (system, role, feature). In this paper, we combine Triadic Concept Analysis (TCA) and Large Language Model (LLM) prompting to suggest the user-story set required to develop a new system relying on the variability logic of an existing system family. This process consists in 1) computing 3-dimensional variability expressed as a set of TCA implications, 2) providing the designer with intelligible design options, 3) capturing the designer's selection of options, 4) proposing a first user-story set corresponding to this selection, 5) consolidating its validity according to the implications identified in step 1, while completing it if necessary, and 6) leveraging LLM to have a more comprehensive website. This process is evaluated with a dataset comprising the user-story sets of 67 similar-purpose websites.

cs.SE↗

arXiv2025

Synergizing Self-Regulation and Artificial-Intelligence Literacy Towards Future Human-AI Integrative Learning

Self-regulated learning (SRL) and Artificial-Intelligence (AI) literacy are becoming key competencies for successful human-AI interactive learning, vital to future education. However, despite their importance, students face imbalanced and underdeveloped SRL and AI literacy capabilities, inhibiting effective using AI for learning. This study analyzed data from 1,704 Chinese undergraduates using clustering methods to uncover four learner groups reflecting developing process(Potential, Development, Master, and AI-Inclined) characterized by varying SRL and AI literacy differentiation. Results highlight obvious disparities in SRL and AI literacy synchronization, with the Master Group achieving balanced development and critical AI-using for SRL, while AI-Inclined Group demonstrate over-reliance on AI and poor SRL application. The Potential Group showed a close mutual promotion trend between SRL and AI literacy, while the Development Group showed a discrete correlation. Resources and instructional guidance support emerged as key factors affecting these differentiations. To translate students to master SRL-AI literacy level and progress within it, the study proposes differentiated support strategies and suggestions. Synergizing SRL and AI literacy growth is the core of development, ensuring equitable and advanced human-centered interactive learning models for future human-AI integrating.

cs.CY↗