arXiv subjects

Sihan Chen

Publications and source records attributed to Sihan Chen.

At least 37 records · Page 2Linked to original sources

arXiv2026

Digital Twin-Enabled Mobility-Aware Cooperative Caching in Vehicular Edge Computing

With the advancement of vehicle-to-vehicle (V2V) ad hoc networks and wireless communication technologies, mobile edge caching has become a key enabler for enhancing network performance and user experience. However, traditional federated learning-based collaborative caching approaches in vehicular scenarios suffer from inadequate client selection mechanisms and limited prediction accuracy, which result in suboptimal cache hit ratios and increased content transmission latency. To address these challenges, we propose a Digital Twin-based Asynchronous Federated Learning-driven Predictive Edge Caching with Deep Reinforcement Learning (DAPR) framework. DAPR employs an intelligent client selection strategy based on asynchronous federated learning, which leverages mobility prediction and data quality assessment to avoid selecting highly mobile clients or clients with low-quality data, thereby significantly improving model convergence efficiency. In addition, we design a GRU-VAE prediction model that uses a Variational Autoencoder (VAE) to capture latent data distribution features and Gated Recurrent Units (GRUs) to model temporal dependencies, thereby substantially enhancing the accuracy of content request prediction. The predicted content popularities are then fed into a deep reinforcement learning-driven caching decision engine to dynamically optimize edge caching resource allocation. Extensive experiments demonstrate that DAPR achieves superior performance in terms of average reward, cache hit ratio, and transmission latency, thereby effectively improving the overall efficiency of vehicular edge caching systems.

cs.NI↗

arXiv2026

Quantizing Bosonized Fermi Surfaces

Bosonization describes Fermi surface dynamics in terms of a collective field that lives on a part of phase space. While sensible semiclassically, the challenge of treating such a field quantum mechanically has prevented bosonization from providing as powerful a nonperturbative tool as in one dimension. We show that general Fermi surfaces can be exactly described by a particular $N\to \infty$ limit of a $U(N)_1$ WZW model, with a tower of irrelevant corrections. This matrix-valued description encodes the noncommutative nature of phase space, and its (solvable) strongly coupled dynamics resolves the naive overcounting of degrees of freedom of the collective field without the need to cut the Fermi surface into patches. This approach furthermore provides a quantitative tool to systematically study power-law corrections to Fermi surface dynamics.

cond-mat.str-el↗

arXiv2026

WINA: Weight Informed Neuron Activation for Accelerating Large Language Model Inference

The growing computational demands of large language models (LLMs) make efficient inference and activation strategies increasingly critical. While recent approaches, such as Mixture-of-Experts (MoE), leverage selective activation but require specialized training, training-free sparse activation methods offer broader applicability and superior resource efficiency through their plug-and-play design. However, many existing methods rely solely on hidden state magnitudes to determine activation, resulting in high approximation errors and suboptimal inference accuracy. To address these limitations, we propose WINA (Weight Informed Neuron Activation), a novel, simple, and training-free sparse activation framework that jointly considers hidden state magnitudes and the column-wise $\ell_2$-norms of weight matrices. We show that this leads to a sparsification strategy that obtains optimal approximation error bounds with theoretical guarantees tighter than existing techniques. Empirically, WINA also outperforms state-of-the-art methods (e.g., TEAL) by up to $2.94\%$ in average performance at the same sparsity levels, across a diverse set of LLM architectures and datasets. These results position WINA as a new performance frontier for training-free sparse activation in LLM inference, advancing training-free sparse activation methods and setting a robust baseline for efficient inference. The source code is available at https://github.com/microsoft/wina.

cs.LG↗

arXiv2026

StableQAT: Stable Quantization-Aware Training at Ultra-Low Bitwidths

Quantization-aware training (QAT) is essential for deploying large models under strict memory and latency constraints, yet achieving stable and robust optimization at ultra-low bitwidths remains challenging. Common approaches based on the straight-through estimator (STE) or soft quantizers often suffer from gradient mismatch, instability, or high computational overhead. As such, we propose StableQAT, a unified and efficient QAT framework that stabilizes training in ultra low-bit settings via a novel, lightweight, and theoretically grounded surrogate for backpropagation derived from a discrete Fourier analysis of the rounding operator. StableQAT strictly generalizes STE as the latter arises as a special case of our more expressive surrogate family, yielding smooth, bounded, and inexpensive gradients that improve QAT training performance and stability across various hyperparameter choices. In experiments, StableQAT exhibits stable and efficient QAT at 2-4 bit regimes, demonstrating improved training stability, robustness, and superior performance with negligible training overhead against standard QAT techniques. Our code is available at https://github.com/microsoft/StableQAT.

cs.LG↗

arXiv2025

Nonreciprocal Transport with Quantum Geometric Origin in Layered Hybrid Perovskite

Quantum geometry quantifies how the electron wavefunction evolves distinctly from conventional transport theory. In noncentrosymmetric materials, nonreciprocal transport with quantum geometric origin remains prominent with localized charge independent of vanished group velocity. The discovery of such nonreciprocal and nonlinear responses was realized by recent advances in two-dimensional materials. As a promising candidate, the electronic structure and symmetry of layered hybrid perovskites can be deliberately designed and manipulated by incorporating selected organic ligands. Despite the observation of exotic photogalvanic effects and chiral optical effects, the underlying mechanism how these nonlinear responses are enabled in the multi-quantum well structures remained unclear. Here we demonstrated the quantum geometric origin for interlayer spontaneous photocurrent in (PEA)2PbI4. Contrary to assumptions that charge transport across the 2D planes is limited, we observed a spontaneous photocurrent along this crystalline orientation. Theoretical analysis using a tight-binding model identifies shift current as the microscopic origin. This quantum geometric effect is enabled by ionic displacements from centrosymmetric coordinates and enhanced by multiband transition high-density bands of the layered hybrid crystal. We anticipate that such unique low-dimensional systems with structure can provide fertile ground for discovering novel optoelectronic functionalities.

cond-mat.mtrl-sci↗

arXiv2025

Enhancing Hole Mobility in Monolayer $WSe_{2}$ p-FETs via Process-Induced Compression

Understanding the interactions between strain, interfacial mechanics, and electrical performance is critical for designing beyond silicon electronics based on hetero-integrated 2D materials. Through combined experiment and simulation, we demonstrated and analyzed the enhancement of hole mobility in p-type monolayer $WSe_{2}$ field effect transistors (FETs) under biaxial compression. We tracked FET performance versus strain by incrementing compressive strain to $WSe_{2}$ channels via sequential AlOx deposition and performing intermediate photoluminescence and transport measurements. The hole mobility factor increased at a rate of 340 $\pm$ 95 %/%$ε$, and the on-current factor increased at a rate of 460 $\pm$ 340 %/%$ε$. Simulation revealed that the enhancement under compression arises primarily from a reduction in inter-valley scattering between the $Γ$--K valence bands, and the rate is robust against variations in carrier density, impurity density, or dielectric environment. These findings show that compressive strain is a powerful technique for enhancing performance in 2D p-FETs and that it is multiplicative with defect and doping engineering.

cond-mat.mtrl-sci↗

arXiv2025

CogniPair: From LLM Chatbots to Conscious AI Agents -- GNWT-Based Multi-Agent Digital Twins for Social Pairing -- Dating & Hiring Applications

Current large language model (LLM) agents lack authentic human psychological processes necessary for genuine digital twins and social AI applications. To address this limitation, we present a computational implementation of Global Workspace Theory (GNWT) that integrates human cognitive architecture principles into LLM agents, creating specialized sub-agents for emotion, memory, social norms, planning, and goal-tracking coordinated through a global workspace mechanism. However, authentic digital twins require accurate personality initialization. We therefore develop a novel adventure-based personality test that evaluates true personality through behavioral choices within interactive scenarios, bypassing self-presentation bias found in traditional assessments. Building on these innovations, our CogniPair platform enables digital twins to engage in realistic simulated dating interactions and job interviews before real encounters, providing bidirectional cultural fit assessment for both romantic compatibility and workplace matching. Validation using 551 GNWT-Agents and Columbia University Speed Dating dataset demonstrates 72% correlation with human attraction patterns, 77.8% match prediction accuracy, and 74% agreement in human validation studies. This work advances psychological authenticity in LLM agents and establishes a foundation for intelligent dating platforms and HR technology solutions.

cs.AI↗

arXiv2025

ExpertLongBench: Benchmarking Language Models on Expert-Level Long-Form Generation Tasks with Structured Checklists

This paper introduces ExpertLongBench, an expert-level benchmark containing 11 tasks from 9 domains that reflect realistic expert workflows and applications. Beyond question answering, the application-driven tasks in ExpertLongBench demand long-form outputs that can exceed 5,000 tokens and strict adherence to domain-specific requirements. Notably, each task in ExpertLongBench includes a rubric, designed or validated by domain experts, to specify task requirements and guide output evaluation. Furthermore, we propose CLEAR, an evaluation framework that supports accurate evaluation of long-form model outputs in our benchmark. To achieve fine-grained, expert-aligned evaluation, CLEAR derives checklists from both model outputs and references by extracting information corresponding to items in the task-specific rubric. Checklist items of model outputs are then compared with corresponding items of reference outputs to assess their correctness, enabling grounded evaluation. We benchmark 13 popular large language models (LLMs) and analyze components in CLEAR, showing that (1) existing LLMs, with the top performer Gemini-2.5-Pro achieving only a 33.4 F1 score, require significant improvement for expert-level tasks; (2) models can generate content corresponding to the required aspects, but far from correct; and (3) accurate checklist extraction and comparison in CLEAR can be achieved by open-weight models for more scalable, reproducible, and low-cost usage.

cs.CL↗

arXiv2025

SymRTLO: Enhancing RTL Code Optimization with LLMs and Neuron-Inspired Symbolic Reasoning

Optimizing Register Transfer Level (RTL) code is crucial for improving the power, performance, and area (PPA) of digital circuits in the early stages of synthesis. Manual rewriting, guided by synthesis feedback, can yield high-quality results but is time-consuming and error-prone. Most existing compiler-based approaches have difficulty handling complex design constraints. Large Language Model (LLM)-based methods have emerged as a promising alternative to address these challenges. However, LLM-based approaches often face difficulties in ensuring alignment between the generated code and the provided prompts. This paper presents SymRTLO, a novel neuron-symbolic RTL optimization framework that seamlessly integrates LLM-based code rewriting with symbolic reasoning techniques. Our method incorporates a retrieval-augmented generation (RAG) system of optimization rules and Abstract Syntax Tree (AST)-based templates, enabling LLM-based rewriting that maintains syntactic correctness while minimizing undesired circuit behaviors. A symbolic module is proposed for analyzing and optimizing finite state machine (FSM) logic, allowing fine-grained state merging and partial specification handling beyond the scope of pattern-based compilers. Furthermore, a fast verification pipeline, combining formal equivalence checks with test-driven validation, further reduces the complexity of verification. Experiments on the RTL-Rewriter benchmark with Synopsys Design Compiler and Yosys show that SymRTLO improves power, performance, and area (PPA) by up to 43.9%, 62.5%, and 51.1%, respectively, compared to the state-of-the-art methods.

cs.AR↗

arXiv2025

Active assembly and non-reciprocal dynamics of elastic membranes

Equilibrium self-assembly and conventional materials processing techniques fall far short of mimicking dynamic self-actuating processes that are commonplace throughout biology. To bridge the gap between living and synthetic matter, we study adhesive non-thermal fibers immersed in an active fluid. Autonomous chaotic flows power non-equilibrium fiber dynamics, inducing their collisions, generating connections, and weaving a membrane-shaped elastic network. This active assembly generates a hierarchy of shapes, structures, and dynamical processes spanning nanometers to centimeters. Ultimately, it generates an active membrane that exhibits global limit cycles induced by a non-reciprocal coupling between the elastic membrane deformations and the alignment axis of the polar active fluid. Our work merges self-assembly with active matter, demonstrating self-processing materials wherein hierarchical life-like structures and dynamics emerge from an initially structureless suspension.

cond-mat.soft↗

arXiv2025

PersonaTwin: A Multi-Tier Prompt Conditioning Framework for Generating and Evaluating Personalized Digital Twins

While large language models (LLMs) afford new possibilities for user modeling and approximation of human behaviors, they often fail to capture the multidimensional nuances of individual users. In this work, we introduce PersonaTwin, a multi-tier prompt conditioning framework that builds adaptive digital twins by integrating demographic, behavioral, and psychometric data. Using a comprehensive data set in the healthcare context of more than 8,500 individuals, we systematically benchmark PersonaTwin against standard LLM outputs, and our rigorous evaluation unites state-of-the-art text similarity metrics with dedicated demographic parity assessments, ensuring that generated responses remain accurate and unbiased. Experimental results show that our framework produces simulation fidelity on par with oracle settings. Moreover, downstream models trained on persona-twins approximate models trained on individuals in terms of prediction and fairness metrics across both GPT-4o-based and Llama-based models. Together, these findings underscore the potential for LLM digital twin-based approaches in producing realistic and emotionally nuanced user simulations, offering a powerful tool for personalized digital user modeling and behavior analysis.

cs.CL↗

arXiv2025

Robust Maximum $L_q$-Likelihood Covariance Estimation for Replicated Spatial Data

Parameter estimation with the maximum $L_q$-likelihood estimator (ML$q$E) is an alternative to the maximum likelihood estimator (MLE) that considers the $q$-th power of the likelihood values for some $q<1$. In this method, extreme values are down-weighted because of their lower likelihood values, which yields robust estimates. In this work, we study the properties of the ML$q$E for spatial data with replicates. We investigate the asymptotic properties of the ML$q$E for Gaussian random fields with a Matérn covariance function, and carry out simulation studies to investigate the numerical performance of the ML$q$E. We show that it can provide more robust and stable estimation results when some of the replicates in the spatial data contain outliers. In addition, we develop a mechanism to find the optimal choice of the hyper-parameter $q$ for the ML$q$E. The robustness of our approach is further verified on a United States precipitation dataset. Compared with other robust methods for spatial data, our proposal is more intuitive and easier to understand, yet it performs well when dealing with datasets containing outliers.

stat.ME↗

arXiv2025

Chirality across scales in tissue dynamics

Chiral processes that lack mirror symmetry pervade nature from enantioselective molecular interactions to the asymmetric development of organisms. An outstanding challenge at the interface between physics and biology consists in bridging the multiple scales between microscopic and macroscopic chirality. Here, we combine theory, experiments and modern inference algorithms to study a paradigmatic example of dynamic chirality transfer across scales: the generation of tissue-scale flows from subcellular forces. The distinctive properties of our microscopic graph model and the corresponding coarse-grained viscoelasticity are that (i) net cell proliferation is spatially inhomogeneous and (ii) cellular dynamics cannot be expressed as an energy gradient. To overcome the general challenge of inferring microscopic model parameters from noisy high-dimensional data, we develop a nudged automatic differentiation algorithm (NADA) that can handle large fluctuations in cell positions observed in single tissue snapshots. This data-calibrated microscopic model quantitatively captures proliferation-driven tissue flows observed at large scales in our experiments on fibroblastoma cell cultures. Beyond chirality, our inference algorithm can be used to extract interpretable graph models from limited amounts of noisy data of living and inanimate cellular systems such as networks of convection cells and flowing foams.

cond-mat.soft↗

arXiv2025

A Fully Generative Motivational Interviewing Counsellor Chatbot for Moving Smokers Towards the Decision to Quit

The conversational capabilities of Large Language Models (LLMs) suggest that they may be able to perform as automated talk therapists. It is crucial to know if these systems would be effective and adhere to known standards. We present a counsellor chatbot that focuses on motivating tobacco smokers to quit smoking. It uses a state-of-the-art LLM and a widely applied therapeutic approach called Motivational Interviewing (MI), and was evolved in collaboration with clinician-scientists with expertise in MI. We also describe and validate an automated assessment of both the chatbot's adherence to MI and client responses. The chatbot was tested on 106 participants, and their confidence that they could succeed in quitting smoking was measured before the conversation and one week later. Participants' confidence increased by an average of 1.7 on a 0-10 scale. The automated assessment of the chatbot showed adherence to MI standards in 98% of utterances, higher than human counsellors. The chatbot scored well on a participant-reported metric of perceived empathy but lower than typical human counsellors. Furthermore, participants' language indicated a good level of motivation to change, a key goal in MI. These results suggest that the automation of talk therapy with a modern LLM has promise.

cs.CL↗

arXiv2025

An LLM-based Self-Evolving Security Framework for 6G Space-Air-Ground Integrated Networks

Recently emerged 6G space-air-ground integrated networks (SAGINs), which integrate satellites, aerial networks, and terrestrial communications, offer ubiquitous coverage for various mobile applications. However, the highly dynamic, open, and heterogeneous nature of SAGINs poses severe security issues. Forming a defense line of SAGINs suffers from two preliminary challenges: 1) accurately understanding massive unstructured multi-dimensional threat information to generate defense strategies against various malicious attacks, 2) rapidly adapting to potential unknown threats to yield more effective security strategies. To tackle the above two challenges, we propose a novel security framework for SAGINs based on Large Language Models (LLMs), which consists of two key ingredients LLM-6GNG and 6G-INST. Our proposed LLM-6GNG leverages refined chain-of-thought (CoT) reasoning and dynamic multi-agent mechanisms to analyze massive unstructured multi-dimensional threat data and generate comprehensive security strategies, thus addressing the first challenge. Our proposed 6G-INST relies on a novel self-evolving method to automatically update LLM-6GNG, enabling it to accommodate unknown threats under dynamic communication environments, thereby addressing the second challenge. Additionally, we prototype the proposed framework with ns-3, OpenAirInterface (OAI), and software-defined radio (SDR). Experiments on three benchmarks demonstrate the effectiveness of our framework. The results show that our framework produces highly accurate security strategies that remain robust against a variety of unknown attacks. We will release our code to contribute to the community.

cs.CR↗

arXiv2025

VALOR: Vision-Audio-Language Omni-Perception Pretraining Model and Dataset

In this paper, we propose a Vision-Audio-Language Omni-peRception pretraining model (VALOR) for multi-modal understanding and generation. Different from widely-studied vision-language pretraining models, VALOR jointly models relationships of vision, audio and language in an end-to-end manner. It contains three separate encoders for single modality representations, and a decoder for multimodal conditional text generation. We design two pretext tasks to pretrain VALOR model, including Multimodal Grouping Alignment (MGA) and Multimodal Grouping Captioning (MGC). MGA projects vision, language and audio to the same common space, building vision-language, audio-language and audiovisual-language alignment simultaneously. MGC learns how to generate text tokens in conditions of vision, audio or their both. To promote vision-audio-language pretraining research, we construct a large-scale high-quality tri-modality dataset named VALOR-1M, which contains 1M audiable videos with human annotated audiovisual captions. Extensive experiments show that VALOR can learn strong multimodal correlations and be generalized to various downstream tasks (e.g., retrieval, captioning and question answering), with different input modalities (e.g., vision-language, audio-language and audiovisual-language). VALOR achieves new state-of-the-art performances on series of public cross-modality benchmarks. Code and data are available at project page https://casia-iva-group.github.io/projects/VALOR.

cs.LG↗

arXiv2024

PyOD 2: A Python Library for Outlier Detection with LLM-powered Model Selection

Outlier detection (OD), also known as anomaly detection, is a critical machine learning (ML) task with applications in fraud detection, network intrusion detection, clickstream analysis, recommendation systems, and social network moderation. Among open-source libraries for outlier detection, the Python Outlier Detection (PyOD) library is the most widely adopted, with over 8,500 GitHub stars, 25 million downloads, and diverse industry usage. However, PyOD currently faces three limitations: (1) insufficient coverage of modern deep learning algorithms, (2) fragmented implementations across PyTorch and TensorFlow, and (3) no automated model selection, making it hard for non-experts. To address these issues, we present PyOD Version 2 (PyOD 2), which integrates 12 state-of-the-art deep learning models into a unified PyTorch framework and introduces a large language model (LLM)-based pipeline for automated OD model selection. These improvements simplify OD workflows, provide access to 45 algorithms, and deliver robust performance on various datasets. In this paper, we demonstrate how PyOD 2 streamlines the deployment and automation of OD models and sets a new standard in both research and industry. PyOD 2 is accessible at [https://github.com/yzhao062/pyod](https://github.com/yzhao062/pyod). This study aligns with the Web Mining and Content Analysis track, addressing topics such as the robustness of Web mining methods and the quality of algorithmically-generated Web data.

cs.LG↗

arXiv2024

Semantic Entropy Can Simultaneously Benefit Transmission Efficiency and Channel Security of Wireless Semantic Communications

Recently proliferated deep learning-based semantic communications (DLSC) focus on how transmitted symbols efficiently convey a desired meaning to the destination. However, the sensitivity of neural models and the openness of wireless channels cause the DLSC system to be extremely fragile to various malicious attacks. This inspires us to ask a question: "Can we further exploit the advantages of transmission efficiency in wireless semantic communications while also alleviating its security disadvantages?". Keeping this in mind, we propose SemEntropy, a novel method that answers the above question by exploring the semantics of data for both adaptive transmission and physical layer encryption. Specifically, we first introduce semantic entropy, which indicates the expectation of various semantic scores regarding the transmission goal of the DLSC. Equipped with such semantic entropy, we can dynamically assign informative semantics to Orthogonal Frequency Division Multiplexing (OFDM) subcarriers with better channel conditions in a fine-grained manner. We also use the entropy to guide semantic key generation to safeguard communications over open wireless channels. By doing so, both transmission efficiency and channel security can be simultaneously improved. Extensive experiments over various benchmarks show the effectiveness of the proposed SemEntropy. We discuss the reason why our proposed method benefits secure transmission of DLSC, and also give some interesting findings, e.g., SemEntropy can keep the semantic accuracy remain 95% with 60% less transmission.

cs.CR↗