arXiv subjects

Yue Wu

Publications and source records attributed to Yue Wu.

At least 127 records · Page 7Linked to original sources

arXiv2025

TreeBoN: Enhancing Inference-Time Alignment with Speculative Tree-Search and Best-of-N Sampling

Inference-time alignment enhances the performance of large language models without requiring additional training or fine-tuning but presents challenges due to balancing computational efficiency with high-quality output. Best-of-N (BoN) sampling, as a simple yet powerful approach, generates multiple responses and selects the best one, achieving improved performance but with a high computational cost. We propose TreeBoN, a novel framework that integrates a speculative tree-search strategy into Best-of-N (BoN) Sampling. TreeBoN maintains a set of parent nodes, iteratively branching and pruning low-quality responses, thereby reducing computational overhead while maintaining high output quality. Our approach also leverages token-level rewards from Direct Preference Optimization (DPO) to guide tree expansion and prune low-quality paths. We evaluate TreeBoN using AlpacaFarm, HH-RLHF, UltraFeedback, GSM8K, and TutorEval datasets, demonstrating consistent improvements. Specifically, TreeBoN achieves the highest win rate of 65% on TutorEval and around 60% win rates across other different datasets, outperforming standard BoN with the same computational cost and showcasing its scalability and alignment efficacy.

cs.CL↗

arXiv2025

Nearly Optimal Approximation Rates for Deep Super ReLU Networks on Sobolev Spaces

This paper introduces deep super ReLU networks (DSRNs) as a method for approximating functions in Sobolev spaces measured by Sobolev norms $W^{m,p}$ for $m\in\mathbb{N}$ with $m\ge 2$ and $1\le p\le +\infty$. Standard ReLU deep neural networks (ReLU DNNs) cannot achieve this goal. DSRNs consist primarily of ReLU DNNs, and several layers of the square of ReLU added at the end to smooth the networks output. This approach retains the advantages of ReLU DNNs, leading to the straightforward training. The paper also proves the optimality of DSRNs by estimating the VC-dimension of higher-order derivatives of DNNs, and obtains the generalization error in Sobolev spaces via an estimate of the pseudo-dimension of higher-order derivatives of DNNs.

math.NA↗

arXiv2025

Prompt-aligned Gradient for Prompt Tuning

Thanks to the large pre-trained vision-language models (VLMs) like CLIP, we can craft a zero-shot classifier by "prompt", e.g., the confidence score of an image being "[CLASS]" can be obtained by using the VLM provided similarity measure between the image and the prompt sentence "a photo of a [CLASS]". Therefore, prompt shows a great potential for fast adaptation of VLMs to downstream tasks if we fine-tune the prompt-based similarity measure. However, we find a common failure that improper fine-tuning may not only undermine the prompt's inherent prediction for the task-related classes, but also for other classes in the VLM vocabulary. Existing methods still address this problem by using traditional anti-overfitting techniques such as early stopping and data augmentation, which lack a principled solution specific to prompt. We present Prompt-aligned Gradient, dubbed ProGrad, to prevent prompt tuning from forgetting the the general knowledge learned from VLMs. In particular, ProGrad only updates the prompt whose gradient is aligned (or non-conflicting) to the "general direction", which is represented as the gradient of the KL loss of the pre-defined prompt prediction. Extensive experiments demonstrate the stronger few-shot generalization ability of ProGrad over state-of-the-art prompt tuning methods. Codes are available at https://github.com/BeierZhu/Prompt-align.

cs.CV↗

arXiv2025

Excising dead components in the surface code using minimally invasive alterations: A performance study

The physical implementation of a large-scale error-corrected quantum processor will necessarily need to mitigate the presence of defective (thereby "dead") physical components in its operation, for example, identified during bring-up of the device or detected in the middle of a computation. In the context of solid-state qubits, the quantum error correcting protocol operating in the presence of dead components should ideally (i) use the same native operation set as that without dead components, (ii) maximize salvaging of functional components, and (iii) use a consistent global operating schedule which optimizes logical qubit performance and is compatible with the control requirements of the system. The scheme proposed by Grans-Samuelsson et al. [Quantum 8, 1429 (2024)] satisfies all three of these criteria: it effectively excises (cuts out) dead components from the surface code using minimally invasive alterations (MIA). We conduct extensive numerical simulations of this proposal for the pairwise-measurement-based surface code protocol in the presence of dead components under circuit-level noise. To that end, we also describe techniques to automatically construct performant check (detector) bases directly from circuits without manual circuit annotation, which may be of independent interest. Both the MIA scheme and this automated check basis computation can be readily used with measurement-based as well as CNOT-based circuits, and the results presented here demonstrate state-of-the-art performance.

quant-ph↗

arXiv2025

VRBench: A Benchmark for Multi-Step Reasoning in Long Narrative Videos

We present VRBench, the first long narrative video benchmark crafted for evaluating large models' multi-step reasoning capabilities, addressing limitations in existing evaluations that overlook temporal reasoning and procedural validity. It comprises 960 long videos (with an average duration of 1.6 hours), along with 8,243 human-labeled multi-step question-answering pairs and 25,106 reasoning steps with timestamps. These videos are curated via a multi-stage filtering process including expert inter-rater reviewing to prioritize plot coherence. We develop a human-AI collaborative framework that generates coherent reasoning chains, each requiring multiple temporally grounded steps, spanning seven types (e.g., event attribution, implicit inference). VRBench designs a multi-phase evaluation pipeline that assesses models at both the outcome and process levels. Apart from the MCQs for the final results, we propose a progress-level LLM-guided scoring metric to evaluate the quality of the reasoning chain from multiple dimensions comprehensively. Through extensive evaluations of 12 LLMs and 19 VLMs on VRBench, we undertake a thorough analysis and provide valuable insights that advance the field of multi-step reasoning.

cs.CV↗

arXiv2025

On the Fast-radio-burst-associated X-ray Bursts: Inverse Compton Scattering of Radio Photons by an Extreme Pair Flow During Magnetosphere Activities

The Galactic fast radio burst (FRB) FRB 200428 was associated with a short X-ray burst (XRB) from the magnetar SGR J1935+2154 during one of its active phases. This FRB-associated XRB exhibits distinct properties compared to other typical XRBs, including a significantly higher cutoff energy and a steeper power-law index. Its recovered X-ray light curve shows a multiple-peak structure, with the time of arrival offset from that of the FRB. These unique features imply a special physical link between the FRB and X-ray emissions. In 2022 October, a similar FRB-XRB association was detected from the same source. In this paper, we propose a model in which the observed spectral and temporal features of the associated XRBs can be attributed to the inverse Compton scattering (ICS) of FRB photons by an extreme pair flow around the light cylinder, with a bulk Lorentz factor of $Γ\sim10$ and a power-law distribution in the comoving frame, characterized by a typical Lorentz factor $γ^\prime_\mathrm{m}\sim5\times 10^4$. This extreme pair flow could originate from the compression of a transient pulse of $\sim10^{40}-10^{41}\mathrm{erg\,s^{-1}}$ and the acceleration through magnetic reconnection in the current sheet during magnetar activity. The Doppler-boosted ICS spectra and the arrival time shifts in such a scenario can well explain the observed features of the FRB 200428-associated XRB and can also account for another associated event in 2022.

astro-ph.HE↗

arXiv2025

Randomised Euler-Maruyama Method for SDEs with Hölder Continuous Drift Coefficient Driven by $α$-stable Lévy Process

In this paper, we examine the performance of randomised Euler-Maruyama (EM) method for additive time-inhomogeneous SDEs with an irregular drift driven by symmetric $α$-table process, $α\in (1,2)$. In particular, the drift is assumed to be $β$-Hölder continuous in time and bounded $η$-Hölder continuous in space with $β,η\in (0,1]$. The strong order of convergence of the randomised EM in $L^p$-norm is shown to be $1/2+(β\wedge (η/α)\wedge(1/2))-\varepsilon$ for an arbitrary $\varepsilon\in (0,1/2)$, higher than the one of standard EM, which cannot exceed $β$. The result for the case of $α\in (1,2)$ extends the almost optimal order of convergence of randomised EM obtained in (arXiv:2501.15527) for SDEs driven by Gaussian noise ($α=2$), and coincides with the performance of EM method in simulating time-homogenous SDEs driven by $α$-stable process considered in (arXiv:2208.10052). Various experiments are presented to validate the theoretical performance.

math.PR↗

arXiv2025

Smart Grid: Cyber Attacks, Critical Defense Approaches, and Digital Twin

As a national critical infrastructure, the smart grid has attracted widespread attention for its cybersecurity issues. The development towards an intelligent, digital, and Internet-connected smart grid has attracted external adversaries for malicious activities. It is necessary to enhance its cybersecurity by both improving the existing defense approaches and introducing novel developed technologies to the smart grid context. As an emerging technology, digital twin (DT) is considered as an enabler for enhanced security. However, the practical implementation is quite challenging. This is due to the knowledge barriers among smart grid designers, security experts, and DT developers. Each single domain is a complicated system covering various components and technologies. As a result, works are needed to sort out relevant contents so that DT can be better embedded in the security architecture design of smart grid. In order to meet this demand, our paper covers the above three domains, i.e., smart grid, cybersecurity, and DT. Specifically, the paper i) introduces the background of the smart grid; ii) reviews external cyber attacks from attack incidents and attack methods; iii) introduces critical defense approaches in industrial cyber systems, which include device identification, vulnerability discovery, intrusion detection systems (IDSs), honeypots, attribution, and threat intelligence (TI); iv) reviews the relevant content of DT, including its basic concepts, applications in the smart grid, and how DT enhances the security. In the end, the paper puts forward our security considerations on the future development of DT-based smart grid. The survey is expected to help developers break knowledge barriers among smart grid, cybersecurity, and DT, and provide guidelines for future security design of DT-based smart grid.

cs.CR↗

arXiv2025

A Data-Driven Prescribed-Time Control Framework via Koopman Operator and Adaptive Backstepping

Achieving rapid and time-deterministic stabilization for complex systems characterized by strong nonlinearities and parametric uncertainties presents a significant challenge. Traditional model-based control relies on precise system models, whereas purely data-driven methods often lack formal stability guarantees, limiting their applicability in safety-critical systems. This paper proposes a novel control framework that synergistically integrates data-driven modeling with model-based control. The framework first employs the Extended Dynamic Mode Decomposition with Control (EDMDc) to identify a high-dimensional Koopman linear model and quantify its bounded uncertainty from data. Subsequently, a novel Prescribed-Time Adaptive Backstepping (PTAB) controller is synthesized based on this data-driven model. The design leverages the structural advantages of Koopman linearization to systematically handle model errors and circumvent the "explosion of complexity" issue inherent in traditional backstepping. The proposed controller is validated through simulations on the classic Van der Pol oscillator. The results demonstrate that the controller can precisely stabilize the system states to a small neighborhood of the origin within a user-prescribed time, regardless of the initial conditions, while ensuring the boundedness of all closed-loop signals. This research successfully combines the flexibility of data-driven approaches with the rigor of Lyapunov-based analysis. It provides a high-performance control strategy with quantifiable performance and pre-assignable settling time for nonlinear systems, showcasing its great potential for controlling complex dynamics.

eess.SY↗

arXiv2025

Event Data Association via Robust Model Fitting for Event-based Object Tracking

Event-based approaches, which are based on bio-inspired asynchronous event cameras, have achieved promising performance on various computer vision tasks. However, the study of the fundamental event data association problem is still in its infancy. In this paper, we propose a novel Event Data Association (called EDA) approach to explicitly address the event association and fusion problem. The proposed EDA seeks for event trajectories that best fit the event data, in order to perform unifying data association and information fusion. In EDA, we first asynchronously fuse the event data based on its information entropy. Then, we introduce a deterministic model hypothesis generation strategy, which effectively generates model hypotheses from the fused events, to represent the corresponding event trajectories. After that, we present a two-stage weighting algorithm, which robustly weighs and selects true models from the generated model hypotheses, through multi-structural geometric model fitting. Meanwhile, we also propose an adaptive model selection strategy to automatically determine the number of the true models. Finally, we use the selected true models to associate and fuse the event data, without being affected by sensor noise and irrelevant structures. We evaluate the performance of the proposed EDA on the object tracking task. The experimental results show the effectiveness of EDA under challenging scenarios, such as high speed, motion blur, and high dynamic range conditions.

cs.CV↗

arXiv2025

SimLOB: Learning Representations of Limited Order Book for Financial Market Simulation

Financial market simulation (FMS) serves as a promising tool for understanding market anomalies and the underlying trading behaviors. To ensure high-fidelity simulations, it is crucial to calibrate the FMS model for generating data closely resembling the observed market data. Previous efforts primarily focused on calibrating the mid-price data, leading to essential information loss of the market activities and thus biasing the calibrated model. The Limit Order Book (LOB) data is the fundamental data fully capturing the market micro-structure and is adopted by worldwide exchanges. However, LOB is not applicable to existing calibration objective functions due to its tabular structure not suitable for the vectorized input requirement. This paper proposes to explicitly learn the vectorized representations of LOB with a Transformer-based autoencoder. Then the latent vector, which captures the major information of LOB, can be applied for calibration. Extensive experiments show that the learned latent representation not only preserves the non-linear auto-correlation in the temporal axis, but the precedence between successive price levels of LOB. Besides, it is verified that the performance of the representation learning stage is consistent with the downstream calibration tasks. Thus, this work also progresses the FMS on LOB data, for the first time.

cs.CE↗

arXiv2025

Beyond Bradley-Terry Models: A General Preference Model for Language Model Alignment

Modeling human preferences is crucial for aligning foundation models with human values. Traditional reward modeling methods, such as the Bradley-Terry (BT) reward model, fall short in expressiveness, particularly in addressing intransitive preferences. In this paper, we introduce preference embedding, an approach that embeds responses into a latent space to capture intricate preference structures efficiently, achieving linear query complexity. Additionally, we propose preference score-based General Preference Optimization (GPO), which generalizes reward-based reinforcement learning from human feedback (RLHF). Experimental results show that our General Preference embedding Model (GPM) consistently outperforms the BT reward model on the RewardBench benchmark and effectively models cyclic preferences where any BT reward model behaves like a random guess. Furthermore, evaluations on downstream tasks such as AlpacaEval2.0, following the language model post-training with GPO and our general preference model, reveal performance improvements over BT models. These findings indicate that our method may enhance the alignment of foundation models with nuanced human values. The code is available at https://github.com/general-preference/general-preference-model.

cs.AI↗

arXiv2025

Online Learning Control Strategies for Industrial Processes with Application for Loosening and Conditioning

This paper proposes a novel adaptive Koopman Model Predictive Control (MPC) framework, termed HPC-AK-MPC, designed to address the dual challenges of time-varying dynamics and safe operation in complex industrial processes. The framework integrates two core strategies: online learning and historically-informed safety constraints. To contend with process time-variance, a Recursive Extended Dynamic Mode Decomposition (rEDMDc) technique is employed to construct an adaptive Koopman model capable of updating its parameters from real-time data, endowing the controller with the ability to continuously learn and track dynamic changes. To tackle the critical issue of safe operation under model uncertainty, we introduce a novel Historical Process Constraint (HPC) mechanism. This mechanism mines successful operational experiences from a historical database and, by coupling them with the confidence level of the online model, generates a dynamic "safety corridor" for the MPC optimization problem. This approach transforms implicit expert knowledge into explicit, adaptive constraints, establishing a dynamic balance between pursuing optimal performance and ensuring robust safety. The proposed HPC-AK-MPC method is applied to a real-world tobacco loosening and conditioning process and systematically validated using an "advisor mode" simulation framework with industrial data. Experimental results demonstrate that, compared to historical operations, the proposed method significantly improves the Process Capability Index (Cpk) for key quality variables across all tested batches, proving its substantial potential in enhancing control performance while guaranteeing operational safety.

eess.SY↗

arXiv2025

Tensor-to-Tensor Models with Fast Iterated Sum Features

Data in the form of images or higher-order tensors is ubiquitous in modern deep learning applications. Owing to their inherent high dimensionality, the need for subquadratic layers processing such data is even more pressing than for sequence data. We propose a novel tensor-to-tensor layer with linear cost in the input size, utilizing the mathematical gadget of ``corner trees'' from the field of permutation counting. In particular, for order-two tensors, we provide an image-to-image layer that can be plugged into image processing pipelines. On the one hand, our method can be seen as a higher-order generalization of state-space models. On the other hand, it is based on a multiparameter generalization of the signature of iterated integrals (or sums). The proposed tensor-to-tensor concept is used to build a neural network layer called the Fast Iterated Sums (FIS) layer which integrates seamlessly with other layer types. We demonstrate the usability of the FIS layer with both classification and anomaly detection tasks. By replacing some layers of a smaller ResNet architecture with FIS, a similar accuracy (with a difference of only 0.1\%) was achieved in comparison to a larger ResNet while reducing the number of trainable parameters and multi-add operations. The FIS layer was also used to build an anomaly detection model that achieved an average AUROC of 97.3\% on the texture images of the popular MVTec AD dataset. The processing and modelling codes are publicly available at https://github.com/diehlj/fast-iterated-sums.

cs.CV↗

arXiv2025

ChatMyopia: An AI Agent for Pre-consultation Education in Primary Eye Care Settings

Large language models (LLMs) show promise for tailored healthcare communication but face challenges in interpretability and multi-task integration particularly for domain-specific needs like myopia, and their real-world effectiveness as patient education tools has yet to be demonstrated. Here, we introduce ChatMyopia, an LLM-based AI agent designed to address text and image-based inquiries related to myopia. To achieve this, ChatMyopia integrates an image classification tool and a retrieval-augmented knowledge base built from literature, expert consensus, and clinical guidelines. Myopic maculopathy grading task, single question examination and human evaluations validated its ability to deliver personalized, accurate, and safe responses to myopia-related inquiries with high scalability and interpretability. In a randomized controlled trial (n=70, NCT06607822), ChatMyopia significantly improved patient satisfaction compared to traditional leaflets, enhancing patient education in accuracy, empathy, disease awareness, and patient-eyecare practitioner communication. These findings highlight ChatMyopia's potential as a valuable supplement to enhance patient education and improve satisfaction with medical services in primary eye care settings.

cs.HC↗

arXiv2025

Neural nanophotonic object detector with ultra-wide field-of-view

Intelligent object detection, which extracts crucial information like targets categories and locations, plays a vital role in emerging technologies including autonomous driving, the Internet of Things, and next-generation mobile communication systems. With the advancement of intelligent object detectors towards higher integration and miniaturization, their portability and adaptability to a broader range of scenarios have been significantly enhanced. However, this progress comes at the cost of reduced detection quality and narrower field-of-view, which severely impacts overall performances. Here we present a neural nanophotonic object detector based on a metalens array, capable of delivering high-quality imaging with an ultra-wide field-of-view of 135°. The combined neural network not only further improves the imaging quality, but also enables the detector to achieve high-precision target recognition and localization. Moreover, we integrated the neural nanophotonic object detector into a miniature unmanned aerial vehicle to enable wide-angle imaging and intelligent recognition of various real-world dynamic objects, demonstrating the high mobility and flexibility of our neural nanophotonic object detector. Our study presents a systematic framework for advancing revolutionary intelligent detection systems, offering significant potential for a wide range of future applications.

physics.optics↗

arXiv2025

Random periodic solutions for stochastic differential equations with non-uniform dissipativity

This paper is concerned with the existence and uniqueness of random periodic solutions for stochastic differential equations (SDEs), where the drift terms involved need not to be uniformly dissipative. On the one hand, via the reflection coupling approach, we investigate the existence of random periodic solutions in the sense of distribution for SDEs without memory, where the drifts are merely dissipative at long distance. On the other hand, via the synchronous coupling strategy, we establish respectively the existence of pathwise random periodic solutions for functional SDEs with a finite time lag and an infinite time lag, in which the drifts are only dissipative on average rather than uniformly dissipative with respect to the time parameters.

math.PR↗

arXiv2025

Alita: Generalist Agent Enabling Scalable Agentic Reasoning with Minimal Predefinition and Maximal Self-Evolution

Recent advances in large language models (LLMs) have enabled agents to autonomously perform complex, open-ended tasks. However, many existing frameworks depend heavily on manually predefined tools and workflows, which hinder their adaptability, scalability, and generalization across domains. In this work, we introduce Alita--a generalist agent designed with the principle of "Simplicity is the ultimate sophistication," enabling scalable agentic reasoning through minimal predefinition and maximal self-evolution. For minimal predefinition, Alita is equipped with only one component for direct problem-solving, making it much simpler and neater than previous approaches that relied heavily on hand-crafted, elaborate tools and workflows. This clean design enhances its potential to generalize to challenging questions, without being limited by tools. For Maximal self-evolution, we enable the creativity of Alita by providing a suite of general-purpose components to autonomously construct, refine, and reuse external capabilities by generating task-related model context protocols (MCPs) from open source, which contributes to scalable agentic reasoning. Notably, Alita achieves 75.15% pass@1 and 87.27% pass@3 accuracy, which is top-ranking among general-purpose agents, on the GAIA benchmark validation dataset, 74.00% and 52.00% pass@1, respectively, on Mathvista and PathVQA, outperforming many agent systems with far greater complexity. More details will be updated at $\href{https://github.com/CharlesQ9/Alita}{https://github.com/CharlesQ9/Alita}$.

cs.AI↗