arXiv subjects

Xinyu Li

Publications and source records attributed to Xinyu Li.

At least 19 recordsLinked to original sources

arXiv2026

PLUME: Parameter-Efficient Personalization of Large Language Models via Low-Rank User Modulation in Shared Subspaces

Personalizing large language models (LLMs) is essential for delivering AI assistance that aligns with individual users' styles, intents, and preferences. While per-user fine-tuning can substantially enhance personalization quality, it introduces significant parameter and storage overhead, limiting scalability to large user populations. We propose PLUME (Personalized Low-Rank Adaptation through User Modulation and Shared Subspace), a lightweight framework that achieves efficient and expressive per-user adaptation by leveraging a shared task-specific subspace. Specifically, PLUME first learns a global task subspace from aggregated user data. Personalization is then achieved by training only a lightweight small square matrix within this subspace, enabling each user to obtain a tailored model while keeping shared components fixed. Cross-layer shared parameters and rank-1 residual terms are further introduced to significantly reduce redundancy while maintaining expressiveness. Experiments on multiple personalized text generation benchmarks demonstrate that PLUME achieves comparable or superior performance to strong baselines, while reducing per-user parameters by over 95%. These results establish shared-subspace modulation with minimal residuals as a scalable and semantically grounded approach to LLM personalization.

cs.AI↗

arXiv2026

Aplaud: Adaptive Personalized Low-Rank Decomposition for User-Specific LLM

In this paper, we study the problem of personalized survey response prediction using fine-tuned large language models (LLMs). This task poses unique challenges: limited per-user training data, scalability of model storage, and the need to exploit shared structure across survey questions. To address these issues, we propose Aplaud (Adaptive Personalized Low-rank and User-specific Nested Decomposition), a lightweight and scalable framework for LLM personalization. Aplaud extends the LoRA paradigm by separating adaptation into a frozen, shared low-rank basis and a compact user-specific correction, augmented with a rank-one residual for finer personalization. To further reduce per-user parameter cost and mitigate overfitting, the correction matrix can be factorized into an even lower-rank form. Empirical results demonstrate that Aplaud achieves efficient, scalable personalization across users while outperforming state-of-the-art LoRA-based personalized LLM approaches in both generalization and inference efficiency.

cs.AI↗

arXiv2026

Rapid Energy Dissipation by Colliding Waves in Strongly Magnetized Plasmas

Rapid dissipation of magnetic energy in highly magnetized environments around neutron stars and black holes is a key open question in high-energy astrophysics. We develop a general kinetic picture of counter-propagating wave collisions in magnetized pair plasmas for arbitrary polarizations and find that magnetic energy can be dissipated on the wave-crossing timescale. The two magnetohydrodynamical conditions on the field invariants, $I_1\equiv B^2-E^2>0$ and $I_2\equiv \mathbf{E}\cdot\mathbf{B}=0$, can be spontaneously violated during the collision. Parallel electric fields develop to screen nonzero $I_2$ with little energy loss, consistent with the evolution described by Force-Free Electrodynamics. When magnetic dominance is lost, strong particle energization is triggered, dissipating magnetic energy on the wave-crossing timescale. This dynamical process yields a rapid dissipation channel of magnetic energy and provides a kinetic pathway to high-energy emission.

astro-ph.HE↗

arXiv2026

From Inspiral to Expansion: The Wake-Driven Torque on Binary Black Holes in Gaseous Medium

Binary black holes (BBHs) in gaseous medium, such as active galactic nucleus (AGN) disks, are important gravitational-wave sources, yet the gas-driven torque that governs their orbital evolution remains to be fully understood. Most existing studies of BBHs in gas often approximate the net torque as the sum of independent dynamical friction (DF) forces exerted on each black hole by its own wake, neglecting the mutual gravitational coupling between the two wakes. We perform three-dimensional hydrodynamic simulations of circular BBHs in a uniform flow and find the torque is determined by the wake-wake interactions. The net torque is controlled by a single parameter $\eta \equiv v_g/v_o$, the ratio of the gas flow velocity to the binary orbital velocity. At small $\eta$, the wakes merge into a single overdense envelope and the time-averaged torque is negative; as $\eta$ increases, the wakes separate and the torque becomes positive. In AGN disks, capture-channel binaries in a disk model naturally produce $\eta$ in the positive-torque regime; the expansion timescale is comparable to or shorter than the disk lifetime, suggesting that gas-driven expansion can compete with gravitational-wave inspiral and suppress the capture-channel merger rate.

astro-ph.HE↗

arXiv2026

Localize-Then-Decide Guarantees for LLM Judgments

Large language models (LLMs) are increasingly used as evaluators to assess output quality and preference alignment, yet providing reliable guarantees of agreement with human judgments remains challenging. Recent work introduces confidence-thresholding methods that provide such guarantees for pairwise comparisons, relying on the assumption that higher estimated confidence implies lower disagreement risk with humans. However, this assumption can break down when the number of candidate responses increases, since distributing probability mass across many alternatives can distort confidence estimates. To address this issue, we propose a Localize-Then-Decide framework. First, conformal prediction localizes a small shortlist that contains the human-preferred response with high probability. Then, a calibrated confidence-based rule selectively chooses a single response from this shortlist or abstains. This design restores the monotonic relationship between confidence and disagreement risk and enables high-probability agreement guarantees. Experiments with multiple candidate sizes across several datasets and judge LLMs demonstrate that our framework consistently achieves higher guarantee success rates and substantially higher coverage than single-stage baselines.

cs.CL↗

arXiv2026

Predictive Simulation of Interphases on Li Metal Surface

Interphases remain the least understood components in advanced batteries. Although their properties dictate whether a new battery chemistry could perform as designed, there has never been a reliable way to predict what an interphase could arise from a new electrolyte system due to the absence of atomistic level knowledge about interphasial formation process. In this work, we attempt to develop a simulation method that can universally predict interphasial chemistries formed on Li metal surface, so that the electrolyte engineering would no longer need lengthy Edisonian approaches. By combining a transferable universal polarizable force field and a universal machine learning force field, we simulate interphasial chemistry across chemically diverse electrolyte formulations, and successfully replicate the experimental observation that fluorinated solvents promote the formation of LiF-rich interphases, whereas interphases of more organic origin arise from conventional carbonate-based electrolytes. By directly capturing these spontaneous interfacial reactions behind these interphasial chemistries, our simulations establish molecular-level relationships between electrolyte chemistry, salt concentration, decomposition pathways, and SEI properties, and opens a route toward universal and high-throughput predictive simulation of interphases that is the foundation for AI-driven electrolyte discoveries.

cond-mat.mtrl-sci↗

arXiv2026

AsymSpec: Efficient Cloud-Edge Speculative Decoding over Asymmetric Networks

Cloud-edge speculative decoding places a lightweight draft model at an edge gateway and a higher-quality target model in the cloud, but inserts communication into every speculative block. Under a constrained uplink, candidate messages may queue while the verifier is idle. Stop-and-wait scheduling leaves edge compute underutilized; optimistic same-request runahead can waste work when a rejection or an unexpected bonus token invalidates dependent drafts. We present AsymSpec, which addresses uplink-gated verification and invalid dependent work with two corresponding mechanisms. Its asymmetric verification protocol keeps the common-path acceptance upload compact and moves richer, rejection-only correction information to the downlink. A total-variation (TV) certificate for the residual distribution determines whether a small target top-K response suffices; if not, the protocol progressively escalates through proposal-based exact recovery before falling back to the full distribution. Its confirmed-prefix pipeline exposes only independent, valid requests to the edge scheduler and lets the cloud re-batch arrived blocks, hiding verification waits when another confirmed-prefix request is ready without using same-request runahead. Across three draft-target pairs, two workloads, and three asymmetric network profiles, our end-to-end evaluation shows that AsymSpec delivers 2.82-28.03$\times$ the output-token throughput of the strongest baseline.

cs.DC↗

arXiv2026

From GenAI Virtual Patient Dialogue Logs to Teacher-Interpretable Process Evidence: A Learning Analytics Study in Higher Education

Medical history taking is a dialogue-based clinical reasoning task in which learners must gather, organise, and integrate patient information while the consultation unfolds. Generative AI-powered virtual patients (GenAI VPs) make repeated history taking practice scalable and preserve full turn by turn dialogue. However, these logs are educationally difficult to use directly. Complete transcripts are too detailed for routine teacher review, whereas final scores obscure whether learners followed up patient cues, checked uncertainty, or used summaries to guide later questioning. This study examined whether coded GenAI VP dialogues can provide teacher-interpretable process evidence of clinical reasoning. We analysed 1{,}030 GenAI VP dialogues from 210 second-year medical learners across five weeks chest-pain cases. Each consultation was teacher-scored using a rubric assessing the full history taking dialogue, and consultations were classified within each week as high- or low-rated using the weekly median score. To explain how rated performance was reflected in the dialogue process, we applied three analytic layers to the same coded dialogue data: behavioural prevalence, local co-occurrence using Epistemic Network Analysis, and sequential transition using Transition Network Analysis. High-rated consultations involved more history taking activity, but differences were not simply about volume. High rated consultations more often connected information gathering and symptom exploration with communication, checking, organisation, and synthesis. Summarising and organising moves more often led to verification or mechanism-oriented follow-up. These findings show how layered analysis of GenAI VP dialogue logs can reveal process patterns associated with high rated history taking and support process-focused feedback in medical education.

cs.CL↗

arXiv2026

EgoRecovery: Acquiring Failure Recovery Ability Through Human Recovery Demonstration

Robust embodied robots should be able to recover from failures and retry tasks in order to operate reliably in unstructured and noisy real-world environments. Achieving this capability requires training policies on data that captures recovery behaviors. However, collecting such data through robot teleoperation is difficult to scale, as it is time-consuming to induce diverse failure states, perform corrective actions, and reset the environment. This challenge is further exacerbated by the high diversity of failure modes, which demands substantially more recovery data than success demonstrations. In this work, we show that egocentric human data capturing failure recovery processes provides a scalable alternative. By efficiently arranging task-level failure configurations and recording short recovery segments, human operators can generate more than 10x as much valid recovery data per hour compared to robot teleoperation under our protocol. To address the embodiment gap between human and robot, we propose EgoRecovery, a co-training framework for learning recovery behavior, where human recovery demonstrations are aligned to a compact corrective-intent space shared with robot data, which captures the timing and magnitude of correction. Only a small number of robot recovery demonstrations are required to connect this intent to executable robot actions. At deployment, a learned recovery gate predicts when correction is needed from robot observations and activates the corrective intent only in recovery states. Experiments on real-world recovery tasks show that EgoRecovery improves success from failure starts over robot-only recovery, direct co-training with human recovery data, and direct intent-transfer baselines.

cs.RO↗

arXiv2026

Notes to Self: Can LLMs Benefit from Experiential Abstractions?

Humans distill experience into reusable abstractions, e.g., strategies and cautionary reminders, and apply them to gradually solve problems more effectively. We study whether Large Language Models (LLMs) can similarly benefit from such experiential abstractions. From LLMs' solution traces on the MATH training set, a stronger teacher or the LLMs themselves extract natural-language abstractions into a retrievable library. We explore two usage modes: (1) inference-time retrieval and (2) reinforcement learning (RL) with abstraction-augmented training prompts. Experiential abstractions improve LLM performance on mathematical and logical reasoning benchmarks. Self-extracted abstractions match teacher-extracted ones, and our abstraction usage framework can transfer to other datasets and models. These findings suggest LLMs can extract and apply experiential abstractions much as humans leverage distilled experience.

cs.CL↗

arXiv2026

WarpagePINN: Thermal Warpage Prediction in Advanced Packaging via a Two-Stage Physics-Informed Neural Networks

Thermal warpage has become a critical issue in advanced packaging, primarily caused by the mismatch in coefficients of thermal expansion (CTE) among heterogeneously integrated materials. However, only a limited number of studies have focused on developing computational methods for coupled thermal-warpage prediction in the chiplet. This paper proposes a two-stage physics-informed neural network (WarpagePINN) framework to compute both temperature profile and warpage deformation of chiplets. The neural networks are trained without relying on labeled datasets generated by conventional simulators. In the first stage, the temperature field is modeled using a Fourier series representation that inherently satisfies boundary conditions, and the network is trained solely through a loss function derived from the governing equation. In the second stage, a multilayer perceptron (MLP) is employed for warpage prediction, utilizing a novel hybrid supervisory strategy to optimize the energy-based loss function instead of residual loss. A parametric WarpagePINN is also developed to quantify uncertainties associated with the CTE. Numerical results show that the proposed WarpagePINN framework achieves excellent agreement with conventional finite element methods, with a mean absolute error (MAE) of 0.2 {\mu}m, while achieving a speedup of approximately 1000 {\times} in CTE parameterization studies.

math.NA↗

arXiv2026

Numerical Approximation for Path-Dependent McKean-Vlasov Control with Non-Asymptotic Error Estimates

Path-dependent McKean--Vlasov (MKV) control models large interacting populations with history-dependent dynamics and costs. This paper develops a unified approximation-and-learning framework for continuous time path-dependent MKV problem under open-loop controls. First, an Euler discretization scheme with piecewise-constant controls is shown to achieve a non-asymptotic error of $O(h^{1/4})$. Second, we establish a discrete dynamic programming principle and prove value equivalence between open-loop and history-dependent feedback controls, enabling optimization on a reduced filtration. Third, an interacting particle system is introduced to approximate the continuous-time value, yielding an overall error bound of $O(h^{1/4}) + O(M^{-\gamma})$ for $M$ particles and an explicitly given $\gamma > 0$. Finally, we propose a fully implementable neural-network policy-gradient method using pathwise features. Numerical experiments, including a path-dependent linear-quadratic benchmark, demonstrate the effectiveness of the algorithm.

math.OC↗

arXiv2026

EBench: Elemental Diagnosis of Generalist Mobile Manipulation Policies

We present EBench, a simulation benchmark that diagnoses generalist mobile manipulation policies beyond a single success-rate scalar. EBench comprises 26 diverse and challenging manipulation tasks annotated along 5 capability dimensions and 4 generalization dimensions. We evaluate state-of-the-art generalist manipulation models including $\pi_0$, $\pi_{0.5}$, XVLA, and InternVLA-A1, and reveal that the models exhibit strikingly different capability profiles: $\pi_{0.5}$ achieves the highest test success rate, the best train--test retention, and the strongest mobile manipulation performance; $\pi_0$ leads on dexterous fixed-base and high-precision tasks; XVLA and InternVLA-A1 exhibit complementary strengths across atomic skills and operating regimes. Beyond capability profiling, EBench analyzes the generalization ability from 4 representative perspectives, identifying the impact of different distribution shift factors. The results reveal strengths and weaknesses of models behind an overall score. We hope this benchmark offers a broad set of diagnostic signals to guide iteration on generalist manipulation models.

cs.RO↗

arXiv2026

Causal Ensemble Agent: Hierarchical Causal Discovery with LLM-guided Expert Reweighting

Causal discovery aims to uncover causal structures from observational data, which is crucial for real-world decision-making. However, different causal discovery algorithms can produce divergent results that conflict with each other, complicating the identification of accurate causal graphs. Traditional approaches rely on numerical values and statistical assumptions, often ignoring rich domain-specific information, such as feature descriptions, which could also help structure learning. While recent works explore using Large Language Models (LLMs) to infer causal relations via direct queries, such methods can be unreliable due to a lack of alignment with the actual data. To address these limitations, we propose Causal Ensemble Agent (CEA), a novel framework that aggregates structural insights from statistical discovery experts across different graph levels via linear opinion pooling, and uses an LLM as a meta-referee to dynamically reweight experts when the aggregated confidence is close to the decision boundary, thereby composing an improved and more complete causal graph. Extensive experiments on both synthetic and real-world datasets demonstrate that CEA achieves the strongest overall performance across a wide range of causal discovery methods, highlighting the effectiveness of using LLMs for meta-analysis in causal discovery.

cs.LG↗

arXiv2026

Traits Run Deeper: Trait-Specific Asymmetric Fusion for Personality Assessment

Personality assessment aims to infer stable traits from dynamic behaviors across modalities like language, voice, and facial expressions. Existing approaches often adopt a uniform multimodal fusion strategy for all personality dimensions, overlooking trait-specific modality preferences and causing cross-modal interference. To address this, we propose Traits Run Deeper, a novel personality assessment framework consisting of three components. First, the Multimodal Foundation Representation (MFR) module constructs personality-oriented inputs and incorporates psychology-informed semantic templates as anchors, enabling foundation models to capture trait-relevant behaviors. Second, the Trait-Specific Modality Fusion (TSMF) module employs an asymmetric fusion mechanism, allowing each dimension to selectively exploit different modality pathways to capture heterogeneous preferences while reducing cross-modal contamination. Third, the Distribution-Calibrated Personality Regression (DCPR) module mitigates label imbalance and central tendency bias through target distribution calibration, improving robustness and stability. Experimental results on the AVI Challenge 2026 validation set show that our framework reduces mean squared error (MSE) by approximately 25% compared with the baseline. Consistent improvements on the official test set demonstrate that our method achieves the best performance and ranks first in the AVI Challenge 2026 Personality Assessment Track. The source code will be made available at [https://github.com/MSA-LMC/TraitsRunDeeper](https://github.com/MSA-LMC/TraitsRunDeeper).

cs.CV↗

arXiv2026

Self-Regulated Learning in Essay Writing: Consistency of Strategies and Impact on Outcomes

Background: Abilities for effective self-regulated learning (SRL) are critical for lifelong learning, particularly during adolescence when these skills consolidate and strongly influence future learning. Their importance has grown with the rise of online and blended education. Yet, little is known about how secondary school students self-regulate in online environments, how their SRL processes and strategies evolve, or how they affect outcomes. In secondary education, understanding these processes can reveal patterns and indicators of learning success, informing the design of online support mechanisms. Evidence from repeated-measures designs remains scarce. Objectives: This study aims to examine how secondary school students enact SRL strategies during online essay writing, how these strategies change over time, and how they relate to learning outcomes. Methods: We analysed metacognition-related trace data collected from secondary students during a two-wave online essay-writing task conducted one week apart in two Colombian schools (N = 93 for session 1, N = 95 for session 2) via a digital learning platform. Using a combination of process mining and unsupervised machine learning techniques, we identified dominant SRL strategies grounded in established SRL processes and examined their stability and association with learning outcomes. Results and conclusions: Three dominant SRL strategies were identified. Results showed variability: many students remained in or shifted to Read first, write next, while none used Write intensively, read selectively in session 2. Although less common, latter strategy was positively associated with learning outcomes.

cs.HC↗

arXiv2026

XFreq-GS: Cross-Frequency Wireless Radiation Field Reconstruction with 3D Gaussian Splatting

Channel modeling is fundamental to the analysis, design, and optimization of wireless communication systems, which, however, accurate wireless channel modeling remains challenging, especially given the increasingly complex wireless environments. As an emerging paradigm, 3D Gaussian Splatting (3DGS)-based channel modeling methods achieve accurate wireless radiation field (WRF) reconstruction and high-fidelity spatial spectrum synthesis. However, existing works only consider a single carrier frequency and fail to adapt to wide-range cross-frequency channels. To address this challenge, we propose XFreq-GS, a cross-frequency Gaussian splatting framework for WRF reconstruction. It employs 3D Gaussian primitives with shared geometry and frequency-adaptive radio frequency (RF) attributes to reconstruct cross-frequency WRF, and synthesizes power angular spectrum (PAS) maps for wireless channel modeling. Experiments show that XFreq-GS outperforms state-of-the-art 3DGS-based methods in PAS synthesis and achieves superior cross-frequency generalization. Code is available at https://github.com/KINGAZ1019/XFreq-GS.

eess.SP↗

arXiv2026

Signature Approach for Contextual Bandits with Nonlinear and Path-dependent Rewards

We study contextual bandits with nonlinear and path-dependent rewards through a novel signature-transform-based approach. Leveraging the universal nonlinearity property of signatures, we approximate continuous path-dependent reward functionals by linear functionals in the signature space. This representation enables the use of efficient linear contextual bandit methods while preserving expressive sequential structure. Building on this framework, we propose \texttt{DisSigUCB}, a signature-based disjoint upper confidence bound (UCB) algorithm. Under boundedness and non-degeneracy assumptions, we prove a high-probability data-dependent sublinear regret bound of order $\tilde{\mathcal O}(\sqrt{(d+m)KT})$ where $d$ is the context dimension and $m$ is the signature feature dimension. Synthetic experiments and numerical applications on temperature sensor monitoring, sleep-stage classification, and hospital nurse staffing demonstrate that \texttt{DisSigUCB} consistently outperforms classical linear and kernelized contextual bandit baselines in nonlinear and path-dependent settings.

cs.LG↗