arXiv subjects

Xin Zhang

Publications and source records attributed to Xin Zhang.

At least 397 records · Page 22Linked to original sources

arXiv2025

LongScape: Advancing Long-Horizon Embodied World Models with Context-Aware MoE

Video-based world models hold significant potential for generating high-quality embodied manipulation data. However, current video generation methods struggle to achieve stable long-horizon generation: classical diffusion-based approaches often suffer from temporal inconsistency and visual drift over multiple rollouts, while autoregressive methods tend to compromise on visual detail. To solve this, we introduce LongScape, a hybrid framework that adaptively combines intra-chunk diffusion denoising with inter-chunk autoregressive causal generation. Our core innovation is an action-guided, variable-length chunking mechanism that partitions video based on the semantic context of robotic actions. This ensures each chunk represents a complete, coherent action, enabling the model to flexibly generate diverse dynamics. We further introduce a Context-aware Mixture-of-Experts (CMoE) framework that adaptively activates specialized experts for each chunk during generation, guaranteeing high visual quality and seamless chunk transitions. Extensive experimental results demonstrate that our method achieves stable and consistent long-horizon generation over extended rollouts. Our code is available at: https://github.com/tsinghua-fib-lab/Longscape.

cs.CV↗

arXiv2025

THCM-CAL: Temporal-Hierarchical Causal Modelling with Conformal Calibration for Clinical Risk Prediction

Automated clinical risk prediction from electronic health records (EHRs) demands modeling both structured diagnostic codes and unstructured narrative notes. However, most prior approaches either handle these modalities separately or rely on simplistic fusion strategies that ignore the directional, hierarchical causal interactions by which narrative observations precipitate diagnoses and propagate risk across admissions. In this paper, we propose THCM-CAL, a Temporal-Hierarchical Causal Model with Conformal Calibration. Our framework constructs a multimodal causal graph where nodes represent clinical entities from two modalities: Textual propositions extracted from notes and ICD codes mapped to textual descriptions. Through hierarchical causal discovery, THCM-CAL infers three clinically grounded interactions: intra-slice same-modality sequencing, intra-slice cross-modality triggers, and inter-slice risk propagation. To enhance prediction reliability, we extend conformal prediction to multi-label ICD coding, calibrating per-code confidence intervals under complex co-occurrences. Experimental results on MIMIC-III and MIMIC-IV demonstrate the superiority of THCM-CAL.

cs.CL↗

arXiv2025

DA-Mamba: Dialogue-aware selective state-space model for multimodal engagement estimation

Human engagement estimation in conversational scenarios is essential for applications such as adaptive tutoring, remote healthcare assessment, and socially aware human--computer interaction. Engagement is a dynamic, multimodal signal conveyed by facial expressions, speech, gestures, and behavioral cues over time. In this work we introduce DA-Mamba, a dialogue-aware multimodal architecture that replaces attention-heavy dialogue encoders with Mamba-based selective state-space processing to achieve linear time and memory complexity while retaining expressive cross-modal reasoning. We design a Mamba dialogue-aware selective state-space model composed of three core modules: a Dialogue-Aware Encoder, and two Mamba-based fusion mechanisms: Modality-Group Fusion and Partner-Group Fusion, these modules achieve expressive dialogue understanding. Extensive experiments on three standard benchmarks (NoXi, NoXi-Add, and MPIIGI) show that DA-Mamba surpasses prior state-of-the-art (SOTA) methods in concordance correlation coefficient (CCC), while reducing training time and peak memory; these gains enable processing much longer sequences and facilitate real-time deployment in resource-constrained, multi-party conversational settings. The source code will be available at: https://github.com/kksssssss-ssda/MMEA.

cs.AI↗

arXiv2025

MAIN: Mutual Alignment Is Necessary for instruction tuning

Instruction tuning has empowered large language models (LLMs) to achieve remarkable performance, yet its success heavily depends on the availability of large-scale, high-quality instruction-response pairs. To meet this demand, various methods have been developed to synthesize data at scale. However, current methods for scaling up data generation often overlook a crucial aspect: the alignment between instructions and responses. We hypothesize that the quality of instruction-response pairs is determined not by the individual quality of each component, but by the degree of mutual alignment. To address this, we propose a Mutual Alignment Framework (MAIN) which enforces coherence between instructions and responses through mutual constraints. We demonstrate that MAIN generalizes well across model architectures and sizes, achieving state-of-the-art performance on LLaMA, Mistral, and Qwen models across diverse benchmarks. This work underscores the critical role of instruction-response alignment in enabling generalizable and high-quality instruction tuning for LLMs. All code is available from our repository.

cs.CL↗

arXiv2025

CCrepairBench: A High-Fidelity Benchmark and Reinforcement Learning Framework for C++ Compilation Repair

The automated repair of C++ compilation errors presents a significant challenge, the resolution of which is critical for developer productivity. Progress in this domain is constrained by two primary factors: the scarcity of large-scale, high-fidelity datasets and the limitations of conventional supervised methods, which often fail to generate semantically correct patches.This paper addresses these gaps by introducing a comprehensive framework with three core contributions. First, we present CCrepair, a novel, large-scale C++ compilation error dataset constructed through a sophisticated generate-and-verify pipeline. Second, we propose a Reinforcement Learning (RL) paradigm guided by a hybrid reward signal, shifting the focus from mere compilability to the semantic quality of the fix. Finally, we establish the robust, two-stage evaluation system providing this signal, centered on an LLM-as-a-Judge whose reliability has been rigorously validated against the collective judgments of a panel of human experts. This integrated approach aligns the training objective with generating high-quality, non-trivial patches that are both syntactically and semantically correct. The effectiveness of our approach was demonstrated experimentally. Our RL-trained Qwen2.5-1.5B-Instruct model achieved performance comparable to a Qwen2.5-14B-Instruct model, validating the efficiency of our training paradigm. Our work provides the research community with a valuable new dataset and a more effective paradigm for training and evaluating robust compilation repair models, paving the way for more practical and reliable automated programming assistants.

cs.AI↗

arXiv2025

Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundary

Large language models (LLMs) have demonstrated remarkable capabilities across a wide range of tasks, yet they often refuse to answer legitimate queries--a phenomenon known as overrefusal. Overrefusal typically stems from over-conservative safety alignment, causing models to treat many reasonable prompts as potentially risky. To systematically understand this issue, we probe and leverage the models' safety decision boundaries to analyze and mitigate overrefusal. Our findings reveal that overrefusal is closely tied to misalignment at these boundary regions, where models struggle to distinguish subtle differences between benign and harmful content. Building on these insights, we present RASS, an automated framework for prompt generation and selection that strategically targets overrefusal prompts near the safety boundary. By harnessing steering vectors in the representation space, RASS efficiently identifies and curates boundary-aligned prompts, enabling more effective and targeted mitigation of overrefusal. This approach not only provides a more precise and interpretable view of model safety decisions but also seamlessly extends to multilingual scenarios. We have explored the safety decision boundaries of various LLMs and construct the MORBench evaluation set to facilitate robust assessment of model safety and helpfulness across multiple languages. Code and datasets are available at https://github.com/Master-PLC/RASS.

cs.AI↗

arXiv2025

Equitable coloring of sparse graphs

An equitable coloring of a graph is a proper coloring where the sizes of any two distinct color classes differ by at most one. The celebrated Chen-Lih-Wu Conjecture (CLWC for short) states that every connected graph $G$ that is neither an odd cycle, a $K_r$, nor a $K_{2m+1,2m+1}$ has an equitable $Δ(G)$-coloring. A graph $G$ is in $\mathcal{G}_{m_1,m_2}$ if for all $H\subseteq G$, $\lVert H \rVert\leq m_1|H|$, and if $H$ is bipartite, then $\lVert H \rVert\leq m_2|H|$. In this paper, we confirm CLWC for all graphs $G$ in $\mathcal{G}_{m_1, m_2}$ provided that $m_1\leq 1.8m_2$ and $Δ(G)\geq \frac{2m_1}{1-β}$, where $β$ is a real root of $2m_2(1-x)(1+x)^2-m_1x(2+x)$. By specializing to the case $m_1 = m_2 = d$, we deduce that every $d$-degenerate graph $G$ with $Δ(G) \geq 6.21d$ admits an equitable $r$-coloring for all $r \geq Δ(G)$, thereby improving the previous best-known lower bound of $10d$ on $Δ(G)$ established by Kostochka and Nakprasit in 2005. A graph is $k$-planar if it can be drawn in the plane so that each edge is crossed at most $k$ times. CLWC had been confirmed for planar graphs $G$ with $Δ(G) \geq 8$ (Kostochka, Lin, and Xiang, 2024) and for $1$-planar graphs $G$ with $Δ(G) \geq 13$ (Cranston and Mahmoud, 2025). As an immediate application of our main result, we extend this confirmation to all $k$-planar graphs $G$ with $k \geq 2$ and $Δ(G) \geq \sqrt{383k}$.

math.CO↗

arXiv2025

OASIS: A Deep Learning Framework for Universal Spectroscopic Analysis Driven by Novel Loss Functions

The proliferation of spectroscopic data across various scientific and engineering fields necessitates automated processing. We introduce OASIS (Omni-purpose Analysis of Spectra via Intelligent Systems), a machine learning (ML) framework for technique-independent, automated spectral analysis, encompassing denoising, baseline correction, and comprehensive peak parameter (location, intensity, FWHM) retrieval without human intervention. OASIS achieves its versatility through models trained on a strategically designed synthetic dataset incorporating features from numerous spectroscopy techniques. Critically, the development of innovative, task-specific loss functions-such as the vicinity peak response (ViPeR) for peak localization-enabled the creation of compact yet highly accurate models from this dataset, validated with experimental data from Raman, UV-vis, and fluorescence spectroscopy. OASIS demonstrates significant potential for applications including in situ experiments, high-throughput optimization, and online monitoring. This study underscores the optimization of the loss function as a key resource-efficient strategy to develop high-performance ML models.

cs.LG↗

arXiv2025

UrQMD Simulations of Higher-order Cumulants in Au+Au Collisions at High Baryon Density

High moments of conserved quantities such as net-baryon, net-electric charge, and net-strangeness in heavy-ion collisions are sensitive to fluctuations caused by the QCD critical point (CP). The event-by-event analysis of high moments of the conserved charges has been widely used in experiments to search for the CP, especially in the RHIC-STAR experiment. In order to establish a {\it dynamical non-critical base line}, especially at the high baryon density region, we have performed a systematic analysis of the proton multiplicity distributions from Au+Au collisions at 3 $\leq$ $\sqrt{s_{NN}}$ $\leq$ 9.2 GeV collisions. The results on beam energy, centrality and rapidity width dependence of proton (factorial) cumulants, up to the $4^{th}$ order, are extracted from the calculations of the hadronic transport model UrQMD. In addition, the effects of initial volume fluctuation is also discussed. These results will be important when we do physics analysis the RHIC beam energy scan (BES) data, especially for the fixed-target data and experimental data from future CBM experiment at FAIR.

nucl-ex↗

arXiv2025

InterFormer: Effective Heterogeneous Interaction Learning for Click-Through Rate Prediction

Click-through rate (CTR) prediction, which predicts the probability of a user clicking an ad, is a fundamental task in recommender systems. The emergence of heterogeneous information, such as user profile and behavior sequences, depicts user interests from different aspects. A mutually beneficial integration of heterogeneous information is the cornerstone towards the success of CTR prediction. However, most of the existing methods suffer from two fundamental limitations, including (1) insufficient inter-mode interaction due to the unidirectional information flow between modes, and (2) aggressive information aggregation caused by early summarization, resulting in excessive information loss. To address the above limitations, we propose a novel module named InterFormer to learn heterogeneous information interaction in an interleaving style. To achieve better interaction learning, InterFormer enables bidirectional information flow for mutually beneficial learning across different modes. To avoid aggressive information aggregation, we retain complete information in each data mode and use a separate bridging arch for effective information selection and summarization. Our proposed InterFormer achieves state-of-the-art performance on three public datasets and a large-scale industrial dataset.

cs.IR↗

arXiv2025

Turnpike properties for zero-sum stochastic linear quadratic differential games of Markovian regime switching system

This paper investigates the long-time behavior of zero-sum stochastic linear-quadratic (SLQ) differential games within Markov regime-switching diffusion systems and establishes the turnpike property of the optimal triple. By verifying the convergence of the associated coupled differential Riccati equations (CDREs) along with their convergence rate, we show that, for a sufficiently large time horizon, the equilibrium strategy in the finite-horizon problem can be closely approximated by that of the infinite-horizon problem. Furthermore, this study enhances and extends existing results concerning zero-sum SLQ differential games over both finite and infinite horizons.

math.OC↗

arXiv2025

Measuring cosmic curvature with non-CMB observations

The cosmic curvature $Ω_{K}$ is an important parameter related to the inflationary cosmology and the ultimate fate of the universe. In this work, we adopt the non-CMB observations to constrain $Ω_{K}$ in the $Λ$CDM model and its extensions. The DESI baryon acoustic oscillation, DES type Ia supernova, cosmic chronometer, and strong gravitational lensing time delay data are considered. We find that the data combination favors an open universe in the $Λ$CDM model, specifically $Ω_{K}=0.108\pm0.056$ at the $1σ$ confidence level, which is in $2.6σ$ tension with the Planck CMB result supporting our universe being slightly closed. In the $Λ$CDM extensions, the data combination is consistent with a spatially flat universe. However, the central value of $Ω_{K}$ is positive and has a significant deviation from zero. We adopt the Akaike information criterion to compare different cosmological models. The result shows that non-flat models fit the observational data better than the flat $Λ$CDM model, which adds evidence to the argument that flat $Λ$CDM is not the ultimate model of cosmology.

astro-ph.CO↗

arXiv2025

CoT-RAG: Integrating Chain of Thought and Retrieval-Augmented Generation to Enhance Reasoning in Large Language Models

Chain-of-thought (CoT) reasoning boosts large language models' (LLMs) performance on complex tasks but faces two key limitations: a lack of reliability when solely relying on LLM-generated reasoning chains and lower reasoning performance from natural language prompts compared with code prompts. To address these issues, we propose CoT-RAG, a novel reasoning framework with three key designs: (i) Knowledge Graph-driven CoT Generation, featuring knowledge graphs to modulate reasoning chain generation of LLMs, thereby enhancing reasoning credibility; (ii) Learnable Knowledge Case-aware RAG, which incorporates retrieval-augmented generation (RAG) into knowledge graphs to retrieve relevant sub-cases and sub-descriptions, providing LLMs with learnable information; (iii) Pseudo Program Prompting Execution, which promotes greater logical rigor by guiding LLMs to execute reasoning tasks as pseudo-programs. Evaluations on nine public datasets spanning three reasoning tasks reveal significant accuracy gains-ranging from 4.0% to 44.3%-over state-of-the-art methods. Furthermore, tests on four domain-specific datasets demonstrate exceptional accuracy and efficient execution, underscoring its practical applicability and scalability. Our code and data are available at https: //github.com/hustlfy123/CoT-RAG.

cs.CL↗

arXiv2025

Analytical modeling of the one-dimensional power spectrum of 21-cm forest based on a halo model method

The 21-cm forest, composed of spectral absorption features from high-redshift background radio sources, provides a unique probe for studying small-scale structures during the epoch of reionization. It is particularly sensitive to detecting small-scale structures and early heating processes. Despite the rich information contained in the 21-cm forest signal, the complexity of directly modeling the signal has led to a lack of effective analytical models. However, the one-dimensional (1D) power spectrum of the 21-cm forest contains valuable information about the matter power spectrum, making analytical modeling feasible. This work employs an analytical modeling approach based on the halo model, which links the distribution of matter to dark matter halos, allowing for effective predictions of cosmic structure formation and its impact on the 21-cm signal. By considering various parameter scenarios within the halo model framework, particularly different dark matter particle masses and varying levels of cosmic heating, we can capture the complexities of small-scale structures and make the 1D power spectrum modeling applicable across a wide range of parameters. This method not only enhances our understanding of the 21-cm forest signal but also provides theoretical support for future observational data. Observing the 21-cm forest with large radio telescopes, such as the Square Kilometre Array, is anticipated to enable simultaneous exploration of dark matter properties and the heating history of the early universe.

astro-ph.CO↗

arXiv2025

AnyGPT: Unified Multimodal LLM with Discrete Sequence Modeling

We introduce AnyGPT, an any-to-any multimodal language model that utilizes discrete representations for the unified processing of various modalities, including speech, text, images, and music. AnyGPT can be trained stably without any alterations to the current large language model (LLM) architecture or training paradigms. Instead, it relies exclusively on data-level preprocessing, facilitating the seamless integration of new modalities into LLMs, akin to the incorporation of new languages. We build a multimodal text-centric dataset for multimodal alignment pre-training. Utilizing generative models, we synthesize the first large-scale any-to-any multimodal instruction dataset. It consists of 108k samples of multi-turn conversations that intricately interweave various modalities, thus equipping the model to handle arbitrary combinations of multimodal inputs and outputs. Experimental results demonstrate that AnyGPT is capable of facilitating any-to-any multimodal conversation while achieving performance comparable to specialized models across all modalities, proving that discrete representations can effectively and conveniently unify multiple modalities within a language model. Demos are shown in https://junzhan2000.github.io/AnyGPT.github.io/

cs.CL↗

arXiv2025

A Literature Review of Literature Reviews in Pattern Analysis and Machine Intelligence

The rapid growth of research in Pattern Analysis and Machine Intelligence (PAMI) has rendered literature reviews essential for consolidating and interpreting knowledge across its many subfields. In this work, we present a comprehensive tertiary analysis of PAMI reviews along three complementary dimensions: (i) identifying structural and statistical regularities in existing surveys; (ii) developing quantitative strategies that help researchers navigate and prioritize within the expanding review corpus; and (iii) critically assessing emerging AI-generated review systems. To support this study, we construct RiPAMI, a large-scale database containing more than 3,000 review articles, and combine narrative synthesis with statistical analysis to capture structural and content-level features. Our analyses reveal distinctive organizational patterns as well as persistent gaps in current review practices. Building on these insights, we propose practical, article-level strategies for indicator-guided navigation that move beyond simple citation counts. Finally, our evaluation of state-of-the-art AI-generated reviews indicates encouraging advances in coherence and organization, yet also highlights enduring weaknesses in reference retrieval, coverage of recent work, and the incorporation of visual elements. Together, these findings provide both a critical appraisal of existing review practices and a forward-looking perspective on how AI-generated reviews can evolve into trustworthy, customizable, and transformative complements to traditional human-authored surveys.

cs.DL↗

arXiv2025

ER-LoRA: Effective-Rank Guided Adaptation for Weather-Generalized Depth Estimation

Monocular depth estimation under adverse weather conditions (e.g.\ rain, fog, snow, and nighttime) remains highly challenging due to the lack of reliable ground truth and the difficulty of learning from unlabeled real-world data. Existing methods often rely on synthetic adverse data with pseudo-labels, which suffer from domain gaps, or employ self-supervised learning, which violates photometric assumptions in adverse scenarios. In this work, we propose to achieve weather-generalized depth estimation by Parameter-Efficient Fine-Tuning (PEFT) of Vision Foundation Models (VFMs), using only a small amount of high-visibility (normal) data. While PEFT has shown strong performance in semantic tasks such as segmentation, it remains underexplored for geometry -- centric tasks like depth estimation -- especially in terms of balancing effective adaptation with the preservation of pretrained knowledge. To this end, we introduce the Selecting-Tuning-Maintaining (STM) strategy, which structurally decomposes the pretrained weights of VFMs based on two kinds of effective ranks (entropy-rank and stable-rank). In the tuning phase, we adaptively select the proper rank number as well as the task-aware singular directions for initialization, based on the entropy-rank and full-tuned weight; while in the maintaining stage, we enforce a principal direction regularization based on the stable-rank. This design guarantees flexible task adaptation while preserving the strong generalization capability of the pretrained VFM. Extensive experiments on four real-world benchmarks across diverse weather conditions demonstrate that STM not only outperforms existing PEFT methods and full fine-tuning but also surpasses methods trained with adverse synthetic data, and even the depth foundation model

cs.CV↗

arXiv2025

Scaling Limits for Exponential Hedging in the Brownian Framework

In this paper, we consider scaling limits of exponential utility indifference prices for European contingent claims in the Bachelier model. We show that the scaling limit can be represented in terms of the \emph{specific relative entropy}, and in addition we construct asymptotic optimal hedging strategies. To prove the upper bound for the limit, we formulate the dual problem as a stochastic control, and show there exists a classical solution to its Hamilton-Jacobi-Bellman (HJB) equation. The proof for the lower bound relies on the duality result for exponential hedging in discrete time.

math.PR↗