arXiv subjects

Yi Li

Publications and source records attributed to Yi Li.

At least 181 records · Page 10Linked to original sources

arXiv2025

Never too Prim to Swim: An LLM-Enhanced RL-based Adaptive S-Surface Controller for AUVs under Extreme Sea Conditions

The adaptivity and maneuvering capabilities of Autonomous Underwater Vehicles (AUVs) have drawn significant attention in oceanic research, due to the unpredictable disturbances and strong coupling among the AUV's degrees of freedom. In this paper, we developed large language model (LLM)-enhanced reinforcement learning (RL)-based adaptive S-surface controller for AUVs. Specifically, LLMs are introduced for the joint optimization of controller parameters and reward functions in RL training. Using multi-modal and structured explicit task feedback, LLMs enable joint adjustments, balance multiple objectives, and enhance task-oriented performance and adaptability. In the proposed controller, the RL policy focuses on upper-level tasks, outputting task-oriented high-level commands that the S-surface controller then converts into control signals, ensuring cancellation of nonlinear effects and unpredictable external disturbances in extreme sea conditions. Under extreme sea conditions involving complex terrain, waves, and currents, the proposed controller demonstrates superior performance and adaptability in high-level tasks such as underwater target tracking and data collection, outperforming traditional PID and SMC controllers.

cs.RO↗

arXiv2025

Toward Fair Graph Neural Networks Via Dual-Teacher Knowledge Distillation

Graph Neural Networks (GNNs) have demonstrated strong performance in graph representation learning across various real-world applications. However, they often produce biased predictions caused by sensitive attributes, such as religion or gender, an issue that has been largely overlooked in existing methods. Recently, numerous studies have focused on reducing biases in GNNs. However, these approaches often rely on training with partial data (e.g., using either node features or graph structure alone), which can enhance fairness but frequently compromises model utility due to the limited utilization of available graph information. To address this tradeoff, we propose an effective strategy to balance fairness and utility in knowledge distillation. Specifically, we introduce FairDTD, a novel Fair representation learning framework built on Dual-Teacher Distillation, leveraging a causal graph model to guide and optimize the design of the distillation process. Specifically, FairDTD employs two fairness-oriented teacher models: a feature teacher and a structure teacher, to facilitate dual distillation, with the student model learning fairness knowledge from the teachers while also leveraging full data to mitigate utility loss. To enhance information transfer, we incorporate graph-level distillation to provide an indirect supplement of graph information during training, as well as a node-specific temperature module to improve the comprehensive transfer of fair knowledge. Experiments on diverse benchmark datasets demonstrate that FairDTD achieves optimal fairness while preserving high model utility, showcasing its effectiveness in fair representation learning for GNNs.

cs.LG↗

arXiv2025

LogPilot: Intent-aware and Scalable Alert Diagnosis for Large-scale Online Service Systems

Effective alert diagnosis is essential for ensuring the reliability of large-scale online service systems. However, on-call engineers are often burdened with manually inspecting massive volumes of logs to identify root causes. While various automated tools have been proposed, they struggle in practice due to alert-agnostic log scoping and the inability to organize complex data effectively for reasoning. To overcome these limitations, we introduce LogPilot, an intent-aware and scalable framework powered by Large Language Models (LLMs) for automated log-based alert diagnosis. LogPilot introduces an intent-aware approach, interpreting the logic in alert definitions (e.g., PromQL) to precisely identify causally related logs and requests. To achieve scalability, it reconstructs each request's execution into a spatiotemporal log chain, clusters similar chains to identify recurring execution patterns, and provides representative samples to the LLMs for diagnosis. This clustering-based approach ensures the input is both rich in diagnostic detail and compact enough to fit within the LLM's context window. Evaluated on real-world alerts from Volcano Engine Cloud, LogPilot improves the usefulness of root cause summarization by 50.34% and exact localization accuracy by 54.79% over state-of-the-art methods. With a diagnosis time under one minute and a cost of only $0.074 per alert, LogPilot has been successfully deployed in production, offering an automated and practical solution for service alert diagnosis.

cs.SE↗

arXiv2025

Cubic Fourfolds with an Order-$7$ Automorphism

We study smooth cubic fourfolds admitting an automorphism of order $7$. It is known that the possible symplectic automorphism groups of such cubic fourfolds are precisely $F_{21}$, $\mathrm{PSL}(2,\mathbb{F}_7)$, and $A_7$. In this paper, we determine all possible full automorphism groups of smooth cubic fourfolds with an automorphism of order $7$. We also investigate the moduli spaces of cubic fourfolds whose automorphism group is either $F_{21}$ or $\mathrm{PSL}(2,\mathbb{F}_7)$, describing them both as GIT quotients and as locally symmetric varieties. In particular, we give an explicit description of the singular cubic fourfolds that appear in the boundary of the corresponding GIT quotients. For these two cases, we determine the commensurability classes of the monodromy groups by explicitly identifying certain arithmetic subgroups. As an interesting consequence, we prove that the period domain for cubic fourfolds equipped with an order-$7$ automorphism is isogenous to a Hilbert modular surface.

math.AG↗

arXiv2025

A Benchmark for Localizing Code and Non-Code Issues in Software Projects

Accurate project localization (e.g., files and functions) for issue resolution is a critical first step in software maintenance. However, existing benchmarks for issue localization, such as SWE-Bench and LocBench, are limited. They focus predominantly on pull-request issues and code locations, ignoring other evidence and non-code files such as commits, comments, configurations, and documentation. To address this gap, we introduce MULocBench, a comprehensive dataset of 1,100 issues from 46 popular GitHub Python projects. Comparing with existing benchmarks, MULocBench offers greater diversity in issue types, root causes, location scopes, and file types, providing a more realistic testbed for evaluation. Using this benchmark, we assess the performance of state-of-the-art localization methods and five LLM-based prompting strategies. Our results reveal significant limitations in current techniques: even at the file level, performance metrics (Acc@5, F1) remain below 40%. This underscores the challenge of generalizing to realistic, multi-faceted issue resolution. To enable future research on project localization for issue resolution, we publicly release MULocBench at https://huggingface.co/datasets/somethingone/MULocBench.

cs.SE↗

arXiv2025

Do Code Semantics Help? A Comprehensive Study on Execution Trace-Based Information for Code Large Language Models

Code Large Language Models (Code LLMs) have opened a new era in programming with their impressive capabilities. However, recent research has revealed critical limitations in their ability to reason about runtime behavior and understand the actual functionality of programs, which poses significant challenges for their post-training and practical deployment. Specifically, Code LLMs encounter two principal issues: (1) a lack of proficiency in reasoning about program execution behavior, as they struggle to interpret what programs actually do during runtime, and (2) the inconsistent and fragmented representation of semantic information, such as execution traces, across existing methods, which hinders their ability to generalize and reason effectively. These challenges underscore the necessity for more systematic approaches to enhance the reasoning capabilities of Code LLMs. To address these issues, we introduce a generic framework to support integrating semantic information~(e.g., execution trace) to code task-relevant prompts, and conduct a comprehensive study to explore the role of semantic information in enhancing the reasoning ability of Code LLMs accordingly. Specifically, we focus on investigating the usefulness of trace-based semantic information in boosting supervised fine-tuning~(SFT) and post-phase inference of Code LLMs. The experimental results surprisingly disagree with previous works and demonstrate that semantic information has limited usefulness for SFT and test time scaling of Code LLM.

cs.SE↗

arXiv2025

A New Class of Mark-Specific Proportional Hazards Models for Recurrent Events: Application to Opioid Refills Among Post-Surgical Patients

Prescription opioids relieve moderate-to-severe pain after surgery, but overprescription can lead to misuse and overdose. Understanding factors associated with post-surgical opioid refills is crucial for improving pain management and reducing opioid-related harms. Conventional methods often fail to account for refill size or dosage and capture patient risk dynamics. We address this gap by treating dosage as a continuously varying mark for each refill event and proposing a new class of mark-specific proportional hazards models for recurrent events. Our marginal model, developed on the gap-time scale with a dual weighting scheme, accommodates event proximity to dosage of interest while accounting for the informative number of recurrences. We establish consistency and asymptotic normality of the estimator and provide a sandwich variance estimator for robust inference. Simulations show improved finite-sample performance over competing methods. We apply the model to data from the Michigan Surgical Quality Collaborative and Michigan Automated Prescription System. Results show that high BMI, smoking, cancer, and open surgery increase hazards of high-dosage refills, while inpatient surgeries elevate refill hazards across all dosages. Black race is associated with higher hazards of low-dosage but lower hazards of high-dosage refills. These findings may inform personalized, dosage-specific pain management strategies.

stat.ME↗

arXiv2025

Magnetic excitations in biaxial-strain detwinned $α$-RuCl$_{3}$

The honeycomb magnet $α$-RuCl$_{3}$ has been a leading candidate for realizing the Kitaev quantum spin liquid (QSL), but its intrinsic spin dynamics have remained obscured by crystal twinning. Here we apply biaxial anisotropic strain to detwin $α$-RuCl$_{3}$ single crystals and directly visualize the intrinsic magnetic excitations using inelastic neutron scattering. We discover that the low-energy spin waves emerge from the $M$ points -- transverse to the magnetic Bragg peaks -- providing direct evidence of anisotropic magnetic interactions in $α$-RuCl$_{3}$. The intrinsic spin-wave spectrum imposes stringent constraints on the extended Kitaev Hamiltonian, yielding a refined, quantitatively consistent set of exchange couplings for the zigzag ground state and its low-energy dynamics. Above the magnon band, we uncover broad excitation continua: while a twofold-symmetric feature near 6 meV at $Γ$ is consistent with bimagnon scattering, the dominant spectral weight forms a sixfold-symmetric continuum extending up to $\sim 16$ meV that cannot be explained by conventional magnons. This strongly supports the presence of fractionalized excitations-a hallmark of Kitaev QSL physics. Our findings establish biaxial strain as a powerful symmetry-breaking probe to access the intrinsic spin dynamics of Kitaev materials and provide critical benchmarks for refining theoretical models of quantum magnetism in $α$-RuCl$_{3}$.

cond-mat.str-el↗

arXiv2025

A large language model system for the field of chemical engineering technology

The development of chemical engineering technology is a multi-stage process that encompasses laboratory research, scaling up, and industrial deployment. This process demands interdisciplinary col laboration and typically incurs significant time and economic costs. To tackle these challenges, we have developed a system based on ChemELLM in this work. This system enables users to interact freely with the chem ical engineering model, establishing a new paradigm for AI-driven in novation and accelerating technological advancements in the chemical sector.If you would like to experience our system, please visit our official website at: https://chemindustry.iflytek.com/chat.

physics.chem-ph↗

arXiv2025

Tuning Magneto-Optical Zero-Reflection via Dual-Channel Hybrid Magnonics

Multi-channel coupling in hybrid systems makes an attractive testbed not only because of the distinct advantages entailed in each constituent mode, but also the opportunity to leverage interference among the various excitation pathways. Here, via combined analytical calculation and experiment, we demonstrate that the phase of the magnetization precession at the interface of a coupled yttrium iron garnet(YIG)/permalloy(Py) bilayer is collectively controlled by the microwave photon field torque and the interlayer exchange torque, manifesting a coherent, dual-channel excitation scheme that effectively tunes the magneto-optic spectrum. The different torque contributions vary with frequency, external bias field, and types of interlayer coupling between YIG and Py, which further results in destructive or constructive interferences between the two excitation channels, and hence, selective suppression or amplification of the hybridized magnon modes.

cond-mat.mtrl-sci↗

arXiv2025

Deep Learning of Semi-Competing Risk Data via a New Neural Expectation-Maximization Algorithm

Prognostication for lung cancer, a leading cause of mortality, remains a complex task, as it needs to quantify the associations of risk factors and health events spanning a patient's entire life. One challenge is that an individual's disease course involves non-terminal (e.g., disease progression) and terminal (e.g., death) events, which form semi-competing relationships. Our motivation comes from the Boston Lung Cancer Study, a large lung cancer survival cohort, which investigates how risk factors influence a patient's disease trajectory. Following developments in the prediction of time-to-event outcomes with neural networks, deep learning has become a focal area for the development of risk prediction methods in survival analysis. However, limited work has been done to predict multi-state or semi-competing risk outcomes, where a patient may experience adverse events such as disease progression prior to death. We propose a novel neural expectation-maximization algorithm to bridge the gap between classical statistical approaches and machine learning. Our algorithm enables estimation of the non-parametric baseline hazards of each state transition, risk functions of predictors, and the degree of dependence among different transitions, via a multi-task deep neural network with transition-specific sub-architectures. We apply our method to the Boston Lung Cancer Study and investigate the impact of clinical and genetic predictors on disease progression and mortality.

stat.ML↗

arXiv2025

Observation of topological switch between Weyl semimetal and third-order topological insulator phases

Weyl semimetals and higher-order topological insulators represent two fundamental yet distinct classes of topological matter. While both have been extensively studied in classical-wave systems, their coexistence and controllable transition within a single platform remain largely unexplored. Meanwhile, implementing three-dimensional spin-orbit couplings, which is crucial for realizing a broad class of higher-dimensional topological phases, continues to pose significant experimental challenges. Here, we experimentally realize three-dimensional spin-orbit couplings and demonstrate that tuning the dimerized spin-orbit coupling strength enables both the coexistence of and a controllable switch between Weyl semimetal and third-order topological insulator phases. By engineering a three-dimensional circuit metamaterial, we synthesize the required spin-orbit interactions and observe hallmark signatures of both phases: frequency spectroscopy reveals the Fermi arcs, while local density of states measurements identify the topological corner modes. Interestingly, the corner mode degeneracy doubles compared to that in the canonical Benalcazar-Bernevig-Hughes model, signaling an enriched topological structure. Our study establishes a fundamental connection between two paradigmatic topological phases and paves the way for further exploring spin-orbit-coupling induced exotic higher-dimensional topological phases.

cond-mat.mes-hall↗

arXiv2025

Stability Optimization and Analysis of Energy Flow Networks versus Different Centrality Measurement

Optimizing the stability and control performance of complex networks often hinges on effectively identifying critical nodes for targeted intervention. Due to their inherent complexity and high dimensionality, large-scale energy flow networks, prevalent in domains like power grids, transportation, and financial systems, present unique challenges in selecting optimal nodes for resource allocation. While numerous centrality measurements, such as Katz centrality, eigenvector centrality, closeness centrality, betweenness centrality, and PageRank, have been proposed to evaluate node importance, the impact of different centrality metrics on stability outcomes remains inadequately understood. Moreover, networks manifest diverse structural characteristics-including small-world, scale-free, and random graph properties-which further complicates the optimization problem. This paper systematically investigates how various node centrality measurements influence control stability across representative complex network structures. A unified energy-flow dynamical model is developed, and performance metrics such as the L1 norm are employed to quantify the network stability implications of employing different centrality metrics. Extensive numerical simulations over statistically generated network ensembles reveal significant variances in stability outcomes, highlighting the crucial role of centrality selection. The findings underscore the sensitivity of energy-flow stability to seemingly minor changes in topological node rankings, providing practical insights for enhancing control efficiency and robustness in real-world networked systems.

physics.soc-ph↗

arXiv2025

Maximal coin-position entanglement and non-Hermitian skin effect in discrete-time quantum walks

A distinctive feature of non-Hermitian systems is the skin effect, which has attracted widespread attention in recent studies. Quantum walks provide a powerful platform for exploring the underlying mechanisms of the non-Hermitian skin effect. Additionally, the generation of hybrid entanglement in quantum walks is recognized as another crucial property. However, the experimentally exploring the influence of skin effect on the evolution of entanglement dynamical in the non-Hermitian system remains a challenge. In this paper, we present a flexible photonic implementation of discrete-time quantum walks over 20 evolution steps using an optimized time-multiplexed loop configuration. Through optimizing the coin parameter, we achieve maximal coin-position entanglement in 20-steps quantum walks. Moreover, we experimentally measure the polarization-averaged growth rates and the evolution of coin-position entanglement for specific coin and loss parameters. We observe the asymmetric Lyapunov exponent profiles and the suppression of entanglement induced by the skin effect in non-Hermitian systems. Interestingly, this entanglement suppression weakens with increasing coin parameters and enhances with increasing loss parameters and evolution steps. Our results demonstrate the potential of quantum walks as a powerful platform for investigating hybrid entanglement properties and skin effect in non-Hermitian systems.

quant-ph↗

arXiv2025

MOC: Meta-Optimized Classifier for Few-Shot Whole Slide Image Classification

Recent advances in histopathology vision-language foundation models (VLFMs) have shown promise in addressing data scarcity for whole slide image (WSI) classification via zero-shot adaptation. However, these methods remain outperformed by conventional multiple instance learning (MIL) approaches trained on large datasets, motivating recent efforts to enhance VLFM-based WSI classification through fewshot learning paradigms. While existing few-shot methods improve diagnostic accuracy with limited annotations, their reliance on conventional classifier designs introduces critical vulnerabilities to data scarcity. To address this problem, we propose a Meta-Optimized Classifier (MOC) comprising two core components: (1) a meta-learner that automatically optimizes a classifier configuration from a mixture of candidate classifiers and (2) a classifier bank housing diverse candidate classifiers to enable a holistic pathological interpretation. Extensive experiments demonstrate that MOC outperforms prior arts in multiple few-shot benchmarks. Notably, on the TCGA-NSCLC benchmark, MOC improves AUC by 10.4% over the state-of-the-art few-shot VLFM-based methods, with gains up to 26.25% under 1-shot conditions, offering a critical advancement for clinical deployments where diagnostic training data is severely limited. Code is available at https://github.com/xmed-lab/MOC.

cs.CV↗

arXiv2025

A15 Phase Ta3Sb Thin Films: Direct Synthesis, Charge Transport and Spin-Orbit Torque

Ta3Sb is one of the A15 compounds that have been predicted to have giant spin Hall conductivities due to the gapped Dirac-like band crossings in their electronic structures. We use co-sputtering to directly synthesize thin films of Ta3Sb and identify a large window of Ta:Sb flux ratio that permits the formation of single-phase A15 structure. These sputtered films have an actual Ta:Sb atomic ratio of 4:1 as determined from Rutherford backscattering spectrometry. Their high resistivity, at the Mott-Ioffe-Regel limit, suggests that the electron mean free path is comparable to interatomic distances. From harmonic Hall and spin-torque ferromagnetic resonance measurements, the intrinsic spin Hall conductivity of thin film Ta3Sb is estimated to be in the range of -526 to -1230 (hbar/e) S/cm at 300 K, lower in magnitude than the predicted value of -1400 (hbar/e) S/cm. First-principles calculations of the electronic structure show that the discrepancy is consistent with an increase of the Fermi level due to the non-ideal stoichiometry needed to stabilize the A15 structure.

cond-mat.mtrl-sci↗

arXiv2025

Calibrated Prediction Set in Fault Detection with Risk Guarantees via Significance Tests

Fault detection is crucial for ensuring the safety and reliability of modern industrial systems. However, a significant scientific challenge is the lack of rigorous risk control and reliable uncertainty quantification in existing diagnostic models, particularly when facing complex scenarios such as distributional shifts. To address this issue, this paper proposes a novel fault detection method that integrates significance testing with the conformal prediction framework to provide formal risk guarantees. The method transforms fault detection into a hypothesis testing task by defining a nonconformity measure based on model residuals. It then leverages a calibration dataset to compute p-values for new samples, which are used to construct prediction sets mathematically guaranteed to contain the true label with a user-specified probability, $1-α$. Fault classification is subsequently performed by analyzing the intersection of the constructed prediction set with predefined normal and fault label sets. Experimental results on cross-domain fault diagnosis tasks validate the theoretical properties of our approach. The proposed method consistently achieves an empirical coverage rate at or above the nominal level ($1-α$), demonstrating robustness even when the underlying point-prediction models perform poorly. Furthermore, the results reveal a controllable trade-off between the user-defined risk level ($α$) and efficiency, where higher risk tolerance leads to smaller average prediction set sizes. This research contributes a theoretically grounded framework for fault detection that enables explicit risk control, enhancing the trustworthiness of diagnostic systems in safety-critical applications and advancing the field from simple point predictions to informative, uncertainty-aware outputs.

cs.AI↗

arXiv2025

LEADRE: Multi-Faceted Knowledge Enhanced LLM Empowered Display Advertisement Recommender System

Display advertising provides significant value to advertisers, publishers, and users. Traditional display advertising systems utilize a multi-stage architecture consisting of retrieval, coarse ranking, and final ranking. However, conventional retrieval methods rely on ID-based learning to rank mechanisms and fail to adequately utilize the content information of ads, which hampers their ability to provide diverse recommendation lists. To address this limitation, we propose leveraging the extensive world knowledge of LLMs. However, three key challenges arise when attempting to maximize the effectiveness of LLMs: "How to capture user interests", "How to bridge the knowledge gap between LLMs and advertising system", and "How to efficiently deploy LLMs". To overcome these challenges, we introduce a novel LLM-based framework called LLM Empowered Display ADvertisement REcommender system (LEADRE). LEADRE consists of three core modules: (1) The Intent-Aware Prompt Engineering introduces multi-faceted knowledge and designs intent-aware pairs that fine-tune LLMs to generate ads tailored to users' personal interests. (2) The Advertising-Specific Knowledge Alignment incorporates auxiliary fine-tuning tasks and Direct Preference Optimization (DPO) to align LLMs with ad semantic and business value. (3) The Efficient System Deployment deploys LEADRE in an online environment by integrating both latency-tolerant and latency-sensitive service. Extensive offline experiments demonstrate the effectiveness of LEADRE and validate the contributions of individual modules. Online A/B test shows that LEADRE leads to a 1.57% and 1.17% GMV lift for serviced users on WeChat Channels and Moments separately. LEADRE has been deployed on both platforms, serving tens of billions of requests each day.

cs.IR↗