arXiv subjects

Yuqing Huang

Publications and source records attributed to Yuqing Huang.

At least 19 recordsLinked to original sources

arXiv2026

A Coin Flip Per Token: Bernoulli Sparse Steering of Large Language Models

Activation steering via sparse autoencoders (SAEs) enables behavioral control of large language models without task-specific fine-tuning, but standard methods apply the steering signal at every generated token, incurring constant per-token perturbation that risks degrading fluency. We ask: is dense intervention necessary? We introduce Stochastic Token Steering (STS), which gates each token independently with probability $p$, and Stochastic Block Steering (SBS), which gates a leading window once per sequence; neither requires a reward model or learned gating policy. Across two model families and two behavioral tasks, steering only 50% of the tokens recovers most of the dense-steering effect while preserving fluency, and steering as few as 30% surpasses prompt-based control. The optimal steering magnitude scales inversely with the intervention ratio, revealing that SAE-mediated control is rate-limited: the behavioral outcome depends on cumulative signal dosage across a sequence.

cs.LG↗

arXiv2026

Harnessing hidden quantum metric response in a 2D magnet via nonlocal photovoltaic effect

The quantum geometry of Bloch wavefunctions underpins a wealth of emergent phenomena in quantum materials. Its imaginary part, the Berry curvature, has long been recognized as a key source for hallmark effects such as quantum Hall and topological phenomena, etc. The real part of quantum geometry, the quantum metric, has recently garnered considerable attention due to predictions of a range of unconventional nonlinear and nonequilibrium responses. Such responses usually vanish in centrosymmetric systems, largely restricting relevant studies to non-centrosymmetric materials. Here we challenge this convention by revealing that the vanished quantum metric response can survive in a hidden form. Using a non-local photovoltaic scheme in a layered magnetic semiconductor, we spatially separate mutually compensating photocurrents and thereby detect such hidden quantum metric response. We demonstrate this effect across distinct magnetic states and down to the ultrathin limit. Moreover, we realize reconfigurable, nonvolatile and probabilistic photodetection enabled by the quantum metric response. These results not only fundamentally expand the material landscape for quantum geometric physics, but also open new gateway to harvest the quantum geometric contributions for state-of-the-art nonvolatile reprogrammable sensing and computing applications.

cond-mat.mes-hall↗

arXiv2026

Enhanced Anomalous Nernst Effect in the Ferromagnetic Kondo Lattice CeCo2As2

The anomalous Nernst effect (ANE), generating a voltage perpendicular to a temperature gradient due to magnetization, is closely linked to the Berry curvature (BC) near the Fermi energy in topological magnets. We report an enhanced spontaneous ANE in the ferromagnetic Kondo lattice CeCo2As2, which features Kondo-screened cerium-based 4f moments embedded in a ferromagnetic d-electron framework. The observed large anomalous Nernst coefficient, greater than the Seebeck coefficient, is attributed to the strong BC present in the f-orbital-dominated flat bands. The enhanced ANE in CeCo2As2 serves as a signature of the Fermi energy pinning within the topological flat band, highlighting the correlation-driven topology in the Kondo lattice.

cond-mat.str-el↗

arXiv2026

Drift-Resilient Temporal Priors for Visual Tracking

Temporal information is crucial for visual tracking, but existing multi-frame trackers are vulnerable to model drift caused by naively aggregating noisy historical predictions. In this paper, we introduce DTPTrack, a lightweight and generalizable module designed to be seamlessly integrated into existing trackers to suppress drift. Our framework consists of two core components: (1) a Temporal Reliability Calibrator (TRC) mechanism that learns to assign a per-frame reliability score to historical states, filtering out noise while anchoring on the ground-truth template; and (2) a Temporal Guidance Synthesizer (TGS) module that synthesizes this calibrated history into a compact set of dynamic temporal priors to provide predictive guidance. To demonstrate its versatility, we integrate DTPTrack into three diverse tracking architectures--OSTrack, ODTrack, and LoRAT-and show consistent, significant performance gains across all baselines. Our best-performing model, built upon an extended LoRATv2 backbone, sets a new state-of-the-art on several benchmarks, achieving a 77.5% Success rate on LaSOT and an 80.3% AO on GOT-10k.

cs.CV↗

arXiv2026

Interactive Tracking: A Human-in-the-Loop Paradigm with Memory-Augmented Adaptation

Existing visual trackers mainly operate in a non-interactive, fire-and-forget manner, making them impractical for real-world scenarios that require human-in-the-loop adaptation. To overcome this limitation, we introduce Interactive Tracking, a new paradigm that allows users to guide the tracker at any time using natural language commands. To support research in this direction, we make three main contributions. First, we present InteractTrack, the first large-scale benchmark for interactive tracking, containing 150 videos with dense bounding box annotations and timestamped language instructions. Second, we propose a comprehensive evaluation protocol and evaluate 25 representative trackers, showing that state-of-the-art methods fail in interactive scenarios; strong performance on conventional benchmarks does not transfer. Third, we introduce Interactive Memory-Augmented Tracking (IMAT), a new baseline that employs a dynamic memory mechanism to learn from user feedback and update tracking behavior accordingly. Our benchmark, protocol, and baseline establish a foundation for developing more intelligent, adaptive, and collaborative tracking systems, bridging the gap between automated perception and human guidance. The full benchmark, tracking results, and analysis are available at https://github.com/NorahGreen/InteractTrack.git.

cs.CV↗

arXiv2026

Cluster-Wise Spatio-Temporal Masking for Efficient Video-Language Pretraining

Large-scale video-language pretraining enables strong generalization across multimodal tasks but often incurs prohibitive computational costs. Although recent advances in masked visual modeling help mitigate this issue, they still suffer from two fundamental limitations: severe visual information loss under high masking ratios and temporal information leakage caused by inter-frame correlations. To address these challenges, we propose ClusterSTM, a Cluster-Wise Spatio-Temporal Masking strategy for efficient video-language pretraining. ClusterSTM first performs intra-frame clustering to partition visual tokens into multiple semantically independent clusters, then conducts cluster-wise masking by retaining the token with the highest temporal density within each cluster. Our masking strategy ensure that the retained tokens capture holistic video content while exhibit strong temporal correlation. Additionally, we introduce a video-text relevance reconstruction objective that aligns high-level multimodal semantics beyond conventional visual reconstruction. Extensive experiments across multiple benchmarks demonstrate that ClusterSTM achieves superior performance on video-text retrieval, video question answering, and video captioning tasks, establishing a new state-of-the-art among efficient video-language models.

cs.CV↗

arXiv2026

Probing quantum geometric nonlinear magnetization via second-harmonic magneto-optical Kerr effect

Quantum geometry provides an intrinsic framework for characterizing the geometric structure of quantum states. It highlights its relevance to various aspects of fundamental physics. However, its direct implications for magnetic phenomena remain largely unexplored. Here, we report the observation of electric-field-induced nonlinear magnetization in the nonmagnetic semimetal WTe$_2$ by using a second-harmonic magneto-optical Kerr effect (SMOKE) spectroscopy. We observe a robust nonlinear SMOKE signal that scales quadratically with current and persists up to 200 K. Theoretical modeling and scaling analysis indicate that this nonlinear magnetization is dominated by the orbital contribution and is intrinsically linked to the quantum Christoffel symbol. Just as the Christoffel symbol is a fundamental quantity encoding spacetime geometry in Einstein's general relativity, our work establishes a direct link between quantum geometry and nonlinear magnetization, and provides a geometric perspective for designing future orbitronic devices.

cond-mat.mes-hall↗

arXiv2025

RAG-IGBench: Innovative Evaluation for RAG-based Interleaved Generation in Open-domain Question Answering

In real-world scenarios, providing user queries with visually enhanced responses can considerably benefit understanding and memory, underscoring the great value of interleaved image-text generation. Despite recent progress, like the visual autoregressive model that unifies text and image processing in a single transformer architecture, generating high-quality interleaved content remains challenging. Moreover, evaluations of these interleaved sequences largely remain underexplored, with existing benchmarks often limited by unimodal metrics that inadequately assess the intricacies of combined image-text outputs. To address these issues, we present RAG-IGBench, a thorough benchmark designed specifically to evaluate the task of Interleaved Generation based on Retrieval-Augmented Generation (RAG-IG) in open-domain question answering. RAG-IG integrates multimodal large language models (MLLMs) with retrieval mechanisms, enabling the models to access external image-text information for generating coherent multimodal content. Distinct from previous datasets, RAG-IGBench draws on the latest publicly available content from social platforms and introduces innovative evaluation metrics that measure the quality of text and images, as well as their consistency. Through extensive experiments with state-of-the-art MLLMs (both open-source and proprietary) on RAG-IGBench, we provide an in-depth analysis examining the capabilities and limitations of these models. Additionally, we validate our evaluation metrics by demonstrating their high correlation with human assessments. Models fine-tuned on RAG-IGBench's training set exhibit improved performance across multiple benchmarks, confirming both the quality and practical utility of our dataset. Our benchmark is available at https://github.com/USTC-StarTeam/RAG-IGBench.

cs.IR↗

arXiv2025

On Predictability of Reinforcement Learning Dynamics for Large Language Models

Recent advances in reasoning capabilities of large language models (LLMs) are largely driven by reinforcement learning (RL), yet the underlying parameter dynamics during RL training remain poorly understood. This work identifies two fundamental properties of RL-induced parameter updates in LLMs: (1) Rank-1 Dominance, where the top singular subspace of the parameter update matrix nearly fully determines reasoning improvements, recovering over 99\% of performance gains; and (2) Rank-1 Linear Dynamics, where this dominant subspace evolves linearly throughout training, enabling accurate prediction from early checkpoints. Extensive experiments across 8 LLMs and 7 algorithms validate the generalizability of these properties. More importantly, based on these findings, we propose AlphaRL, a plug-in acceleration framework that extrapolates the final parameter update using a short early training window, achieving up to 2.5 speedup while retaining \textgreater 96\% of reasoning performance without extra modules or hyperparameter tuning. This positions our finding as a versatile and practical tool for large-scale RL, opening a path toward principled, interpretable, and efficient training paradigm for LLMs.

cs.LG↗

arXiv2025

SelfAug: Mitigating Catastrophic Forgetting in Retrieval-Augmented Generation via Distribution Self-Alignment

Recent advancements in large language models (LLMs) have revolutionized natural language processing through their remarkable capabilities in understanding and executing diverse tasks. While supervised fine-tuning, particularly in Retrieval-Augmented Generation (RAG) scenarios, effectively enhances task-specific performance, it often leads to catastrophic forgetting, where models lose their previously acquired knowledge and general capabilities. Existing solutions either require access to general instruction data or face limitations in preserving the model's original distribution. To overcome these limitations, we propose SelfAug, a self-distribution alignment method that aligns input sequence logits to preserve the model's semantic distribution, thereby mitigating catastrophic forgetting and improving downstream performance. Extensive experiments demonstrate that SelfAug achieves a superior balance between downstream learning and general capability retention. Our comprehensive empirical analysis reveals a direct correlation between distribution shifts and the severity of catastrophic forgetting in RAG scenarios, highlighting how the absence of RAG capabilities in general instruction tuning leads to significant distribution shifts during fine-tuning. Our findings not only advance the understanding of catastrophic forgetting in RAG contexts but also provide a practical solution applicable across diverse fine-tuning scenarios. Our code is publicly available at https://github.com/USTC-StarTeam/SelfAug.

cs.CL↗

arXiv2025

On the Superimposed Noise Accumulation Problem in Sequential Knowledge Editing of Large Language Models

Sequential knowledge editing techniques aim to continuously update knowledge in large language models at low cost, preventing models from generating outdated or incorrect information. However, existing sequential editing methods suffer from a significant decline in editing success rates after long-term editing. Through theoretical analysis and experiments, our findings reveal that as the number of edits increases, the model's output increasingly deviates from the desired target, leading to a drop in editing success rates. We refer to this issue as the superimposed noise accumulation problem. Our further analysis demonstrates that the problem is related to the erroneous activation of irrelevant knowledge and conflicts between activated knowledge. Based on this analysis, a method named DeltaEdit is proposed that reduces conflicts between knowledge through dynamic orthogonal constraint strategies. Experiments show that DeltaEdit significantly reduces superimposed noise, achieving a 16.8% improvement in editing performance over the strongest baseline.

cs.CL↗

arXiv2025

Unconventional bias-dependent tunneling magnetoresistance in van der Waals ferromagnetic/semiconductor heterojunctions

Two-dimensional van der Waals (vdW) ferromagnetic/semiconductor heterojunctions represent an ideal platform for studying and exploiting tunneling magnetoresistance (TMR) effects due to the versatile band structure of semiconductors and their high-quality interfaces. In the all-vdW magnetic tunnel junction (MTJ) devices, both the magnitude and sign of the TMR can be tuned by an applied voltage. Typically, as the bias voltage increases, first the amplitude of the TMR decreases, then the sign of the TMR reverses and/or oscillates. Here, we report on an unconventional bias-dependent TMR in the all-vdW Fe3GaTe2/GaSe/Fe3GaTe2 MTJs, where the TMR first increases, then decreases, and finally undergoes a sign reversal as the bias voltage increases. This dependence cannot be explained by traditional models of MTJs. We propose an in-plane electron momentum (k//) resolved tunneling model that considers both the coherent degree of k// and the decay of the electron wave function through the semiconductor spacer layer. This can explain well the conventional and unconventional bias-dependent TMR. Our results thus provide a deeper understanding of the bias-dependent spin-transport in semiconductor-based MTJs and offer new insights into semiconductor spintronics.

cond-mat.mtrl-sci↗

arXiv2025

Deep Learning within Tabular Data: Foundations, Challenges, Advances and Future Directions

Tabular data remains one of the most prevalent data types across a wide range of real-world applications, yet effective representation learning for this domain poses unique challenges due to its irregular patterns, heterogeneous feature distributions, and complex inter-column dependencies. This survey provides a comprehensive review of state-of-the-art techniques in tabular data representation learning, structured around three foundational design elements: training data, neural architectures, and learning objectives. Unlike prior surveys that focus primarily on either architecture design or learning strategies, we adopt a holistic perspective that emphasizes the universality and robustness of representation learning methods across diverse downstream tasks. We examine recent advances in data augmentation and generation, specialized neural network architectures tailored to tabular data, and innovative learning objectives that enhance representation quality. Additionally, we highlight the growing influence of self-supervised learning and the adaptation of transformer-based foundation models for tabular data. Our review is based on a systematic literature search using rigorous inclusion criteria, encompassing 127 papers published since 2020 in top-tier conferences and journals. Through detailed analysis and comparison, we identify emerging trends, critical gaps, and promising directions for future research, aiming to guide the development of more generalizable and effective tabular data representation methods.

cs.LG↗

arXiv2024

Predicting Stroke through Retinal Graphs and Multimodal Self-supervised Learning

Early identification of stroke is crucial for intervention, requiring reliable models. We proposed an efficient retinal image representation together with clinical information to capture a comprehensive overview of cardiovascular health, leveraging large multimodal datasets for new medical insights. Our approach is one of the first contrastive frameworks that integrates graph and tabular data, using vessel graphs derived from retinal images for efficient representation. This method, combined with multimodal contrastive learning, significantly enhances stroke prediction accuracy by integrating data from multiple sources and using contrastive learning for transfer learning. The self-supervised learning techniques employed allow the model to learn effectively from unlabeled data, reducing the dependency on large annotated datasets. Our framework showed an AUROC improvement of 3.78% from supervised to self-supervised approaches. Additionally, the graph-level representation approach achieved superior performance to image encoders while significantly reducing pre-training and fine-tuning runtimes. These findings indicate that retinal images are a cost-effective method for improving cardiovascular disease predictions and pave the way for future research into retinal and cerebral vessel connections and the use of graph-based retinal vessel representations.

cs.CV↗

arXiv2024

ChemEval: A Comprehensive Multi-Level Chemical Evaluation for Large Language Models

There is a growing interest in the role that LLMs play in chemistry which lead to an increased focus on the development of LLMs benchmarks tailored to chemical domains to assess the performance of LLMs across a spectrum of chemical tasks varying in type and complexity. However, existing benchmarks in this domain fail to adequately meet the specific requirements of chemical research professionals. To this end, we propose \textbf{\textit{ChemEval}}, which provides a comprehensive assessment of the capabilities of LLMs across a wide range of chemical domain tasks. Specifically, ChemEval identified 4 crucial progressive levels in chemistry, assessing 12 dimensions of LLMs across 42 distinct chemical tasks which are informed by open-source data and the data meticulously crafted by chemical experts, ensuring that the tasks have practical value and can effectively evaluate the capabilities of LLMs. In the experiment, we evaluate 12 mainstream LLMs on ChemEval under zero-shot and few-shot learning contexts, which included carefully selected demonstration examples and carefully designed prompts. The results show that while general LLMs like GPT-4 and Claude-3.5 excel in literature understanding and instruction following, they fall short in tasks demanding advanced chemical knowledge. Conversely, specialized LLMs exhibit enhanced chemical competencies, albeit with reduced literary comprehension. This suggests that LLMs have significant potential for enhancement when tackling sophisticated tasks in the field of chemistry. We believe our work will facilitate the exploration of their potential to drive progress in chemistry. Our benchmark and analysis will be available at {\color{blue} \url{https://github.com/USTC-StarTeam/ChemEval}}.

cs.CL↗

arXiv2024

RTracker: Recoverable Tracking via PN Tree Structured Memory

Existing tracking methods mainly focus on learning better target representation or developing more robust prediction models to improve tracking performance. While tracking performance has significantly improved, the target loss issue occurs frequently due to tracking failures, complete occlusion, or out-of-view situations. However, considerably less attention is paid to the self-recovery issue of tracking methods, which is crucial for practical applications. To this end, we propose a recoverable tracking framework, RTracker, that uses a tree-structured memory to dynamically associate a tracker and a detector to enable self-recovery ability. Specifically, we propose a Positive-Negative Tree-structured memory to chronologically store and maintain positive and negative target samples. Upon the PN tree memory, we develop corresponding walking rules for determining the state of the target and define a set of control flows to unite the tracker and the detector in different tracking scenarios. Our core idea is to use the support samples of positive and negative target categories to establish a relative distance-based criterion for a reliable assessment of target loss. The favorable performance in comparison against the state-of-the-art methods on numerous challenging benchmarks demonstrates the effectiveness of the proposed algorithm.

cs.CV↗

arXiv2023

XCloud-VIP: Virtual Peak Enables Highly Accelerated NMR Spectroscopy and Faithful Quantitative Measures

Nuclear Magnetic Resonance (NMR) spectroscopy is an important bio-engineering tool to determine the metabolic concentrations, molecule structures and so on. The data acquisition time, however, is very long in multi-dimensional NMR. To accelerate data acquisition, non-uniformly sampling is an effective way but may encounter severe spectral distortions and unfaithful quantitative measures when the acceleration factor is high. By modelling the acquired signal as the superimposed exponentials, we proposed a virtual peak (VIP) approach to selfadapt the prior spectral information, such as the resonance frequency and peak lineshape, and then feed these information into the reconstruction. The proposed method is further implemented with cloud computing to facilitate online, open, and easy access. Results on simulated and experimental data demonstrate that, compared with the low-rank Hankel matrix method, the new approach reconstructs high-fidelity NMR spectra from highly undersampled data and achieves more accurate quantification. The maximum quantitative errors of distances between nuclear pairs and concentrations of metabolites in mixtures have been reduced by 61.1% and 57.7%, respectively.

physics.med-ph↗

arXiv2023

Non-Uniform Sampling Reconstruction for Symmetrical NMR Spectroscopy by Exploiting Inherent Symmetry

Symmetrical NMR spectroscopy constitutes a vital branch of multidimensional NMR spectroscopy, providing a powerful tool for the structural elucidation of biological macromolecules. Non-Uniform Sampling (NUS) serves as an effective strategy for averting the prohibitive acquisition time of multidimensional NMR spectroscopy by only sampling a few points according to NUS sampling schedules and reconstructing missing points via algorithms. However, current sampling schedules are unable to maintain the accurate recovery of cross peaks that are weak but important. In this work, we propose a novel sampling schedule termed as SCPG (Symmetrical Copy Poisson Gap) and employ CS (Compressed Sensing) methods for reconstruction. We theoretically prove that the symmetrical constraint, apart from sparsity, is implicitly implemented when SCPG is combined with CS methods. The simulated and experimental data substantiate the advantage of SCPG over state-of-the-art 2D Woven PG in the NUS reconstruction of symmetrical NMR spectroscopy.

eess.SP↗