arXiv subjects

Yue Zhao

Publications and source records attributed to Yue Zhao.

At least 181 records · Page 10Linked to original sources

arXiv2025

Dream to Chat: Model-based Reinforcement Learning on Dialogues with User Belief Modeling

World models have been widely utilized in robotics, gaming, and auto-driving. However, their applications on natural language tasks are relatively limited. In this paper, we construct the dialogue world model, which could predict the user's emotion, sentiment, and intention, and future utterances. By defining a POMDP, we argue emotion, sentiment and intention can be modeled as the user belief and solved by maximizing the information bottleneck. By this user belief modeling, we apply the model-based reinforcement learning framework to the dialogue system, and propose a framework called DreamCUB. Experiments show that the pretrained dialogue world model can achieve state-of-the-art performances on emotion classification and sentiment identification, while dialogue quality is also enhanced by joint training of the policy, critic and dialogue world model. Further analysis shows that this manner holds a reasonable exploration-exploitation balance and also transfers well to out-of-domain scenarios such as empathetic dialogues.

cs.CL↗

arXiv2025

SCRA-VQA: Summarized Caption-Rerank for Augmented Large Language Models in Visual Question Answering

Acquiring high-quality knowledge is a central focus in Knowledge-Based Visual Question Answering (KB-VQA). Recent methods use large language models (LLMs) as knowledge engines for answering. These methods generally employ image captions as visual text descriptions to assist LLMs in interpreting images. However, the captions frequently include excessive noise irrelevant to the question, and LLMs generally do not comprehend VQA tasks, limiting their reasoning capabilities. To address this issue, we propose the Summarized Caption-Rerank Augmented VQA (SCRA-VQA), which employs a pre-trained visual language model to convert images into captions. Moreover, SCRA-VQA generates contextual examples for the captions while simultaneously summarizing and reordering them to exclude unrelated information. The caption-rerank process enables LLMs to understand the image information and questions better, thus enhancing the model's reasoning ability and task adaptability without expensive end-to-end training. Based on an LLM with 6.7B parameters, SCRA-VQA performs excellently on two challenging knowledge-based VQA datasets: OK-VQA and A-OKVQA, achieving accuracies of 38.8% and 34.6%. Our code is available at https://github.com/HubuKG/SCRA-VQA.

cs.CV↗

arXiv2025

DAOpt: Modeling and Evaluation of Data-Driven Optimization under Uncertainty with LLMs

Recent advances in large language models (LLMs) have accelerated research on automated optimization modeling. While real-world decision-making is inherently uncertain, most existing work has focused on deterministic optimization with known parameters, leaving the application of LLMs in uncertain settings largely unexplored. To that end, we propose the DAOpt framework including a new dataset OptU, a multi-agent decision-making module, and a simulation environment for evaluating LLMs with a focus on out-of-sample feasibility and robustness. Additionally, we enhance LLMs' modeling capabilities by incorporating few-shot learning with domain knowledge from stochastic and robust optimization.

cs.LG↗

arXiv2025

Beyond the high score: Prosocial ability profiles of multi-agent populations

The development and evaluation of social capabilities in AI agents require complex environments where competitive and cooperative behaviours naturally emerge. While game-theoretic properties can explain why certain teams or agent populations outperform others, more abstract behaviours, such as convention following, are harder to control in training and evaluation settings. The Melting Pot contest is a social AI evaluation suite designed to assess the cooperation capabilities of AI systems. In this paper, we apply a Bayesian approach known as Measurement Layouts to infer the capability profiles of multi-agent systems in the Melting Pot contest. We show that these capability profiles not only predict future performance within the Melting Pot suite but also reveal the underlying prosocial abilities of agents. Our analysis indicates that while higher prosocial capabilities sometimes correlate with better performance, this is not a universal trend-some lower-scoring agents exhibit stronger cooperation abilities. Furthermore, we find that top-performing contest submissions are more likely to achieve high scores in scenarios where prosocial capabilities are not required. These findings, together with reports that the contest winner used a hard-coded solution tailored to specific environments, suggest that at least one top-performing team may have optimised for conditions where cooperation was not necessary, potentially exploiting limitations in the evaluation framework. We provide recommendations for improving the annotation of cooperation demands and propose future research directions to account for biases introduced by different testing environments. Our results demonstrate that Measurement Layouts offer both strong predictive accuracy and actionable insights, contributing to a more transparent and generalisable approach to evaluating AI systems in complex social settings.

cs.AI↗

arXiv2025

Further evidence for natal kick segregation by spectral type in high-mass X-ray binaries

High-mass X-ray binaries (HMXBs) are systems in which a neutron star or black hole accretes material from a massive companion. HMXBs are expected to have experienced a supernova in their evolution. The impulsive kick associated with this event should affect the space velocity of the system in a way that depends on the nature and state of the progenitor binary. Here, we test whether the different evolutionary histories of HMXBs have left a detectable imprint on their peculiar velocities ($V_{\rm pec}$). Using data from Gaia Data Release 3 (Gaia DR3), we first calculate the $V_{\rm pec}$ values for 63 well-known HMXBs hosting a black hole or neutron star and estimate the associated uncertainties via Monte Carlo re-sampling. We then analyse their distribution and check for differences between classes. Overall, $V_{\rm pec}$ estimates extend up to 100 km s$^{-1}$, but with Be/X-ray binaries (BeXRBs) favouring $V_{\rm pec}$ $\lesssim 40$ km s$^{-1}$ and supergiant X-ray binaries (SgXRBs) favouring $V_{\rm pec}$ $\gtrsim 40$ km s$^{-1}$. Based on a Kolmogorov-Smirnov (K-S) test, the null hypothesis that the peculiar velocities of both classes are drawn from the same parent distribution can be robustly rejected, irrespective of the background stellar velocity dispersion. Tests with binary population synthesis demonstrate that SgXRBs typically have shorter orbital periods and higher fractional mass loss than BeXRBs at supernova. We argue that the magnitude of $V_{\rm pec}$ could be used as a complementary feature to distinguish between Be and supergiant systems. These findings extend previous inferences based on two-dimensional kinematics from Hipparcos, and may be explained by the differing nature of the respective progenitors systems between the source classes at the instant of supernova.

astro-ph.HE↗

arXiv2025

Convert Language Model into a Value-based Strategic Planner

Emotional support conversation (ESC) aims to alleviate the emotional distress of individuals through effective conversations. Although large language models (LLMs) have obtained remarkable progress on ESC, most of these studies might not define the diagram from the state model perspective, therefore providing a suboptimal solution for long-term satisfaction. To address such an issue, we leverage the Q-learning on LLMs, and propose a framework called straQ*. Our framework allows a plug-and-play LLM to bootstrap the planning during ESC, determine the optimal strategy based on long-term returns, and finally guide the LLM to response. Substantial experiments on ESC datasets suggest that straQ* outperforms many baselines, including direct inference, self-refine, chain of thought, finetuning, and finite state machines.

cs.CL↗

arXiv2025

A Systematic Survey of Model Extraction Attacks and Defenses: State-of-the-Art and Perspectives

Machine learning (ML) models have significantly grown in complexity and utility, driving advances across multiple domains. However, substantial computational resources and specialized expertise have historically restricted their wide adoption. Machine-Learning-as-a-Service (MLaaS) platforms have addressed these barriers by providing scalable, convenient, and affordable access to sophisticated ML models through user-friendly APIs. While this accessibility promotes widespread use of advanced ML capabilities, it also introduces vulnerabilities exploited through Model Extraction Attacks (MEAs). Recent studies have demonstrated that adversaries can systematically replicate a target model's functionality by interacting with publicly exposed interfaces, posing threats to intellectual property, privacy, and system security. In this paper, we offer a comprehensive survey of MEAs and corresponding defense strategies. We propose a novel taxonomy that classifies MEAs according to attack mechanisms, defense approaches, and computing environments. Our analysis covers various attack techniques, evaluates their effectiveness, and highlights challenges faced by existing defenses, particularly the critical trade-off between preserving model utility and ensuring security. We further assess MEAs within different computing paradigms and discuss their technical, ethical, legal, and societal implications, along with promising directions for future research. This systematic survey aims to serve as a valuable reference for researchers, practitioners, and policymakers engaged in AI security and privacy. Additionally, we maintain an online repository continuously updated with related literature at https://github.com/kzhao5/ModelExtractionPapers.

cs.CR↗

arXiv2025

Score Matching on Large Geometric Graphs for Cosmology Generation

Generative models are a promising tool to produce cosmological simulations but face significant challenges in scalability, physical consistency, and adherence to domain symmetries, limiting their utility as alternatives to $N$-body simulations. To address these limitations, we introduce a score-based generative model with an equivariant graph neural network that simulates gravitational clustering of galaxies across cosmologies starting from an informed prior, respects periodic boundaries, and scales to full galaxy counts in simulations. A novel topology-aware noise schedule, crucial for large geometric graphs, is introduced. The proposed equivariant score-based model successfully generates full-scale cosmological point clouds of up to 600,000 halos, respects periodicity and a uniform prior, and outperforms existing diffusion models in capturing clustering statistics while offering significant computational advantages. This work advances cosmology by introducing a generative model designed to closely resemble the underlying gravitational clustering of structure formation, moving closer to physically realistic and efficient simulators for the evolution of large-scale structures in the universe.

astro-ph.CO↗

arXiv2025

PML method for the stochastic acoustic scattering problem driven by an additive Gaussian noise

This paper is concerned with the time-domain stochastic acoustic scattering problem driven by a spatially white additive Gaussian noise. The main contributions of the work are twofold. First, we prove the existence and uniqueness of the pathwise solution to the scattering problem by applying an abstract Laplace transform inversion theorem. The analysis employs the black box scattering theory to investigate the meromorphic continuation of the Helmholtz resolvent defined on rough fields. Second, based on the piecewise constant approximation of the white noise, we construct an approximate wave solution and establish the error estimate. As a consequence, we develop a PML method and establish the convergence analysis with explicit dependence on the PML layer's thickness and medium properties, as well as the piecewise constant approximation of the white noise.

math.NA↗

arXiv2025

Meta-Learning for Speeding Up Large Model Inference in Decentralized Environments

The deployment of large-scale models, such as large language models (LLMs), incurs substantial costs due to their computational demands. To mitigate these costs and address challenges related to scalability and data security, there is a growing shift towards decentralized systems for model deployment, where choosing efficient inference acceleration schemes become crucial to manage computational resources effectively and enhance system responsiveness. In this work, we address the challenge of selecting optimal acceleration methods in decentralized systems by introducing a meta-learning-based framework. This framework automates the selection process by learning from historical performance data of various acceleration techniques across different tasks. Unlike traditional methods that rely on random selection or expert intuition, our approach systematically identifies the best acceleration strategies based on the specific characteristics of each task. We demonstrate that our meta-learning framework not only streamlines the decision-making process but also consistently outperforms conventional methods in terms of efficiency and performance. Our results highlight the potential of inference acceleration in decentralized AI systems, offering a path towards more democratic and economically feasible artificial intelligence solutions.

cs.LG↗

arXiv2025

Magnetic order dependent photoluminescence from high energy excitons in hBN protected few-layer CrSBr

The detection and manipulation of the spin configurations in layered magnetic semiconductors hold significant interest for developing spintronic devices in two-dimensional limit. In this letter, we report a systematical study on the photoluminescence (PL) from the high energy excitons in few-layer CrSBr and its application on detecting the spin configurations. Besides the broad excitonic emission peak (Xl) at around 1.34 eV, we also observed another strong excitonic emission peak (Xh) at around 1.37 eV in hBN encapsulated 2L sample, which splits into two peaks in 3L and 4L samples. With help of the first principles calculations, we conclude that the Xh peak is associated with the transition between the top valence band and the second lowest conduction band, which is forbidden by the inversion symmetry in 1L CrSBr. Furthermore, the position and intensity of the Xh peak are strongly dependent on the interlayer magnetic order of the CrSBr samples, which provides an efficient way to probe their spin configurations. In addition, when the magnetic field is applied at the easy axis direction, we resolve an intermediate magnetic state besides the antiferromagnetic and ferromagnetic states in 3L and 4L samples. Our results reveal few-layer CrSBr as an ideal platform to study the interaction between the excitons and magnetism.

cond-mat.mes-hall↗

arXiv2025

An inverse moving point source problem in electromagnetics

This paper is concerned with an inverse moving point source problem in electromagnetics. The aim is to reconstruct the moving orbit from the tangential components of magnetic fields taken at a finite number of observation points. The distance function between each observation point and the moving point source is computed by solving a nonlinear ordinary differential equation with an initial value. This ODE system only involves the measurement data from the tangential trace of the magnetic field at observation points. As a consequence, the dynamical measurement data recorded at four non-coplanar points are sufficient to reconstruct the orbit function. A Lipschitz stability is established for the inverse problem, and numerical experiments are reported to demonstrate the effectiveness of the proposed method. Numerical examples have shown that the reconstructed error depends linearly on the noise level and that the wave speed is a critical factor affecting the relative error.

math.NA↗

arXiv2025

JailDAM: Jailbreak Detection with Adaptive Memory for Vision-Language Model

Multimodal large language models (MLLMs) excel in vision-language tasks but also pose significant risks of generating harmful content, particularly through jailbreak attacks. Jailbreak attacks refer to intentional manipulations that bypass safety mechanisms in models, leading to the generation of inappropriate or unsafe content. Detecting such attacks is critical to ensuring the responsible deployment of MLLMs. Existing jailbreak detection methods face three primary challenges: (1) Many rely on model hidden states or gradients, limiting their applicability to white-box models, where the internal workings of the model are accessible; (2) They involve high computational overhead from uncertainty-based analysis, which limits real-time detection, and (3) They require fully labeled harmful datasets, which are often scarce in real-world settings. To address these issues, we introduce a test-time adaptive framework called JAILDAM. Our method leverages a memory-based approach guided by policy-driven unsafe knowledge representations, eliminating the need for explicit exposure to harmful data. By dynamically updating unsafe knowledge during test-time, our framework improves generalization to unseen jailbreak strategies while maintaining efficiency. Experiments on multiple VLM jailbreak benchmarks demonstrate that JAILDAM delivers state-of-the-art performance in harmful content detection, improving both accuracy and speed.

cs.CR↗

arXiv2025

Learning Free Terminal Time Optimal Closed-loop Control of Manipulators

This paper presents a novel approach to learning free terminal time closed-loop control for robotic manipulation tasks, enabling dynamic adjustment of task duration and control inputs to enhance performance. We extend the supervised learning approach, namely solving selected optimal open-loop problems and utilizing them as training data for a policy network, to the free terminal time scenario. Three main challenges are addressed in this extension. First, we introduce a marching scheme that enhances the solution quality and increases the success rate of the open-loop solver by gradually refining time discretization. Second, we extend the QRnet in Nakamura-Zimmerer et al. (2021b) to the free terminal time setting to address discontinuity and improve stability at the terminal state. Third, we present a more automated version of the initial value problem (IVP) enhanced sampling method from previous work (Zhang et al., 2022) to adaptively update the training dataset, significantly improving its quality. By integrating these techniques, we develop a closed-loop policy that operates effectively over a broad domain with varying optimal time durations, achieving near globally optimal total costs.

math.OC↗

arXiv2025

Deep Picard Iteration for High-Dimensional Nonlinear PDEs

We present the Deep Picard Iteration (DPI) method, a new deep learning approach for solving high-dimensional partial differential equations (PDEs). The core innovation of DPI lies in its use of Picard iteration to reformulate the typically complex training objectives of neural network-based PDE solutions into much simpler, standard regression tasks based on function values and gradients. This design not only greatly simplifies the optimization process but also offers the potential for further scalability through parallel data generation. Crucially, to fully realize the benefits of regressing on both function values and gradients in the DPI method, we address the issue of infinite variancein the estimators of gradients by incorporating a control variate, supported by our theoretical analysis. Our experiments on problems up to 100 dimensions demonstrate that DPI consistently outperforms existing state-of-the-art methods, with greater robustness to hyperparameters, particularly in challenging scenarios with long time horizons and strong nonlinearity. The code is available at https://github.com/DeepOptimalControl/DeepPicardIteration.

math.NA↗

arXiv2025

A Survey on Model Extraction Attacks and Defenses for Large Language Models

Model extraction attacks pose significant security threats to deployed language models, potentially compromising intellectual property and user privacy. This survey provides a comprehensive taxonomy of LLM-specific extraction attacks and defenses, categorizing attacks into functionality extraction, training data extraction, and prompt-targeted attacks. We analyze various attack methodologies including API-based knowledge distillation, direct querying, parameter recovery, and prompt stealing techniques that exploit transformer architectures. We then examine defense mechanisms organized into model protection, data privacy protection, and prompt-targeted strategies, evaluating their effectiveness across different deployment scenarios. We propose specialized metrics for evaluating both attack effectiveness and defense performance, addressing the specific challenges of generative language models. Through our analysis, we identify critical limitations in current approaches and propose promising research directions, including integrated attack methodologies and adaptive defense mechanisms that balance security with model utility. This work serves NLP researchers, ML engineers, and security professionals seeking to protect language models in production environments.

cs.CR↗

arXiv2025

PrivacyXray: Detecting Privacy Breaches in LLMs through Semantic Consistency and Probability Certainty

Large Language Models (LLMs) are widely used in sensitive domains, including healthcare, finance, and legal services, raising concerns about potential private information leaks during inference. Privacy extraction attacks, such as jailbreaking, expose vulnerabilities in LLMs by crafting inputs that force the models to output sensitive information. However, these attacks cannot verify whether the extracted private information is accurate, as no public datasets exist for cross-validation, leaving a critical gap in private information detection during inference. To address this, we propose PrivacyXray, a novel framework detecting privacy breaches by analyzing LLM inner states. Our analysis reveals that LLMs exhibit higher semantic coherence and probabilistic certainty when generating correct private outputs. Based on this, PrivacyXray detects privacy breaches using four metrics: intra-layer and inter-layer semantic similarity, token-level and sentence-level probability distributions. PrivacyXray addresses critical challenges in private information detection by overcoming the lack of open-source private datasets and eliminating reliance on external data for validation. It achieves this through the synthesis of realistic private data and a detection mechanism based on the inner states of LLMs. Experiments show that PrivacyXray achieves consistent performance, with an average accuracy of 92.69% across five LLMs. Compared to state-of-the-art methods, PrivacyXray achieves significant improvements, with an average accuracy increase of 20.06%, highlighting its stability and practical utility in real-world applications.

cs.CR↗

arXiv2025

QCD Axion Domain Walls from Super-Cooling First Order Phase Transition

The QCD axion is a well-motivated hypothetical particle beyond the Standard Model (SM) and a compelling dark matter candidate. Its relic abundance is highly sensitive to the thermal history of the universe when the temperature is around the QCD confinement scale. Meanwhile, the NANOGrav Collaboration has reported evidence for a stochastic gravitational wave background, which could originate from a supercooled first-order phase transition (FOPT) with a nucleation temperature around the O(MeV-GeV) scale. We explore how such an FOPT might alter the evolution of the QCD axion. Our findings suggest that it could induce the axion to go through a short stage of mini kinetic misalignment. Moreover, in some parameter regime, the formation of QCD axion domain walls becomes generically expected. This has intriguing implications for both the existence of the QCD axion and the FOPT interpretation of the NANOGrav signal.

hep-ph↗