arXiv subjects

Liu Liu

Publications and source records attributed to Liu Liu.

At least 109 records · Page 6Linked to original sources

arXiv2025

SciAgent: A Unified Multi-Agent System for Generalistic Scientific Reasoning

Recent advances in large language models have enabled AI systems to achieve expert-level performance on domain-specific scientific tasks, yet these systems remain narrow and handcrafted. We introduce SciAgent, a unified multi-agent system designed for generalistic scientific reasoning-the ability to adapt reasoning strategies across disciplines and difficulty levels. SciAgent organizes problem solving as a hierarchical process: a Coordinator Agent interprets each problem's domain and complexity, dynamically orchestrating specialized Worker Systems, each composed of interacting reasoning Sub-agents for symbolic deduction, conceptual modeling, numerical computation, and verification. These agents collaboratively assemble and refine reasoning pipelines tailored to each task. Across mathematics and physics Olympiads (IMO, IMC, IPhO, CPhO), SciAgent consistently attains or surpasses human gold-medalist performance, demonstrating both domain generality and reasoning adaptability. Additionally, SciAgent has been tested on the International Chemistry Olympiad (IChO) and selected problems from the Humanity's Last Exam (HLE) benchmark, further confirming the system's ability to generalize across diverse scientific domains. This work establishes SciAgent as a concrete step toward generalistic scientific intelligence-AI systems capable of coherent, cross-disciplinary reasoning at expert levels.

cs.AI↗

arXiv2025

A Bi-fidelity based asymptotic-preserving neural network for the semiconductor Boltzmann equation and its inverse problem

This paper introduces a Bi-fidelity Asymptotic-Preserving Neural Network (BI-APNNs) framework, designed to efficiently solve forward and inverse problems for the semiconductor Boltzmann equation. Our approach builds upon the Asymptotic-Preserving Neural Network (APNNs) methodology \cite{APNN-transport}, which employs a micro-macro decomposition to handle the model's multiscale nature. We specifically address a key bottleneck in the original APNNs: the slow convergence of the macroscopic density $ρ$ in the near fluid-dynamic regime, i.e., for small Knudsen numbers $\varepsilon$. The core innovation of BI-APNNs is a novel bi-fidelity decomposition of the macroscopic quantity $ρ$, which accurately approximates the true density at small $\varepsilon$, and can be efficiently pre-trained. A separate and more compact neural network is then tasked with learning only the minor correction term, $ρ_{\text{corr}}$. This strategy not only significantly {\it accelerates} the training convergence but also improves the accuracy of the forward problem solution, particularly in the challenging fluid-dynamic limit. Meanwhile, we demonstrate through extensive numerical experiments that our new BI-APNNs yields substantially more accurate and robust results for inverse problems compared to the standard APNNs. Validated on both the semiconductor Boltzmann and the Boltzmann-Poisson systems, our work shows that the bi-fidelity formulation is a powerful enhancement for tackling multiscale kinetic equations, especially when dealing with inverse problems constrained by partial observation data.

math.NA↗

arXiv2025

Remodeling Semantic Relationships in Vision-Language Fine-Tuning

Vision-language fine-tuning has emerged as an efficient paradigm for constructing multimodal foundation models. While textual context often highlights semantic relationships within an image, existing fine-tuning methods typically overlook this information when aligning vision and language, thus leading to suboptimal performance. Toward solving this problem, we propose a method that can improve multimodal alignment and fusion based on both semantics and relationships.Specifically, we first extract multilevel semantic features from different vision encoder to capture more visual cues of the relationships. Then, we learn to project the vision features to group related semantics, among which are more likely to have relationships. Finally, we fuse the visual features with the textual by using inheritable cross-attention, where we globally remove the redundant visual relationships by discarding visual-language feature pairs with low correlation. We evaluate our proposed method on eight foundation models and two downstream tasks, visual question answering and image captioning, and show that it outperforms all existing methods.

cs.CV↗

arXiv2025

Rethinking Pan-sharpening: A New Training Process for Full-Resolution Generalization

The field of pan-sharpening has recently seen a trend towards increasingly large and complex models, often trained on single, specific satellite datasets. This one-dataset, one-model approach leads to high computational overhead and impractical deployment. More critically, it overlooks a core challenge: poor generalization from reduced-resolution (RR) training to real-world full-resolution (FR) data. In response to this issue, we challenge this paradigm. We introduce a multiple-in-one training strategy, where a single, compact model is trained simultaneously on three distinct satellite datasets (WV2, WV3, and GF2). Our experiments show the primary benefit of this unified strategy is a significant and universal boost in FR generalization (QNR) across all tested models, directly addressing this overlooked problem. This paradigm also inherently solves the one-model-per-dataset challenge, and we support it with a highly reproducible, dependency-free codebase for true usability. Finally, we propose PanTiny, a lightweight framework designed specifically for this new, robust paradigm. We demonstrate it achieves a superior performance-to-efficiency balance, proving that principled, simple and robust design is more effective than brute-force scaling in this practical setting. Our work advocates for a community-wide shift towards creating efficient, deployable, and truly generalizable models for pan-sharpening. The code is open-sourced at https://github.com/Zirconium233/PanTiny.

cs.CV↗

arXiv2025

Exploring Category-level Articulated Object Pose Tracking on SE(3) Manifolds

Articulated objects are prevalent in daily life and robotic manipulation tasks. However, compared to rigid objects, pose tracking for articulated objects remains an underexplored problem due to their inherent kinematic constraints. To address these challenges, this work proposes a novel point-pair-based pose tracking framework, termed \textbf{PPF-Tracker}. The proposed framework first performs quasi-canonicalization of point clouds in the SE(3) Lie group space, and then models articulated objects using Point Pair Features (PPF) to predict pose voting parameters by leveraging the invariance properties of SE(3). Finally, semantic information of joint axes is incorporated to impose unified kinematic constraints across all parts of the articulated object. PPF-Tracker is systematically evaluated on both synthetic datasets and real-world scenarios, demonstrating strong generalization across diverse and challenging environments. Experimental results highlight the effectiveness and robustness of PPF-Tracker in multi-frame pose tracking of articulated objects. We believe this work can foster advances in robotics, embodied intelligence, and augmented reality. Codes are available at https://github.com/mengxh20/PPFTracker.

cs.CV↗

arXiv2025

Multi-Task Vehicle Routing Solver via Mixture of Specialized Experts under State-Decomposable MDP

Existing neural methods for multi-task vehicle routing problems (VRPs) typically learn unified solvers to handle multiple constraints simultaneously. However, they often underutilize the compositional structure of VRP variants, each derivable from a common set of basis VRP variants. This critical oversight causes unified solvers to miss out the potential benefits of basis solvers, each specialized for a basis VRP variant. To overcome this limitation, we propose a framework that enables unified solvers to perceive the shared-component nature across VRP variants by proactively reusing basis solvers, while mitigating the exponential growth of trained neural solvers. Specifically, we introduce a State-Decomposable MDP (SDMDP) that reformulates VRPs by expressing the state space as the Cartesian product of basis state spaces associated with basis VRP variants. More crucially, this formulation inherently yields the optimal basis policy for each basis VRP variant. Furthermore, a Latent Space-based SDMDP extension is developed by incorporating both the optimal basis policies and a learnable mixture function to enable the policy reuse in the latent space. Under mild assumptions, this extension provably recovers the optimal unified policy of SDMDP through the mixture function that computes the state embedding as a mapping from the basis state embeddings generated by optimal basis policies. For practical implementation, we introduce the Mixture-of-Specialized-Experts Solver (MoSES), which realizes basis policies through specialized Low-Rank Adaptation (LoRA) experts, and implements the mixture function via an adaptive gating mechanism. Extensive experiments conducted across VRP variants showcase the superiority of MoSES over prior methods.

cs.AI↗

arXiv2025

Deep Learning Based Joint Space-Time-Frequency Domain Channel Prediction for Cell-Free Massive MIMO Systems

The cell-free massive multi-input multi-output (CF-mMIMO) is a promising technology for the six generation (6G) communication systems. Channel prediction will play an important role in obtaining the accurate CSI to improve the performance of CF-mMIMO systems. This paper studies a deep learning (DL) based joint space-time-frequency domain channel prediction for CF-mMIMO. Firstly, the prediction problems are formulated, which can output the multi-step prediction results in parallel without error propagation. Then, a novel channel prediction model is proposed, which adds frequency convolution (FreqConv) and space convolution (SpaceConv) layers to Transformer-encoder. It is able to utilize the space-time-frequency correlations and extract the space correlation in the irregular AP deployment. Next, simulated datasets with different sizes of service areas, UE velocities and scenarios are generated, and correlation analysis and cross-validation are used to determine the optimal hyper-parameters. According to the optimized hyper-parameters, the prediction accuracy and computational complexity are evaluated based on simulated datasets. It is indicated that the prediction accuracy of the proposed model is higher than traditional model, and its computational complexity is lower than traditional Transformer model. After that, the impacts of space-time-frequency correlations on prediction accuracy are studied. Finally, realistic datasets in a high-speed train (HST) long-term evolution (LTE) network are collected to verify the prediction accuracy. The verification results demonstrate that it also achieves higher prediction accuracy compared with traditional models in the HST LTE network.

eess.SP↗

arXiv2025

KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation

Articulated objects, such as laptops and drawers, exhibit significant challenges for 3D reconstruction and pose estimation due to their multi-part geometries and variable joint configurations, which introduce structural diversity across different states. To address these challenges, we propose KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation, a unified framework for reconstructing diverse articulated instances and pose estimation from single view input. Specifically, we first encode complete geometry (SDFs), joint angles, and part segmentation into a structured latent space via a novel Kinematic-Aware VAE (KA-VAE). In addition, we employ two conditional diffusion models: one for regressing global pose (SE(3)) and joint parameters, and another for generating the kinematic-aware latent code from partial observations. Finally, we produce an iterative optimization module that bidirectionally refines reconstruction accuracy and kinematic parameters via Chamfer-distance minimization while preserving articulation constraints. Experimental results on synthetic, semi-synthetic, and real-world datasets demonstrate the effectiveness of our approach in accurately reconstructing articulated objects and estimating their kinematic properties.

cs.CV↗

arXiv2025

MorphoBench: A Benchmark with Difficulty Adaptive to Model Reasoning

With the advancement of powerful large-scale reasoning models, effectively evaluating the reasoning capabilities of these models has become increasingly important. However, existing benchmarks designed to assess the reasoning abilities of large models tend to be limited in scope and lack the flexibility to adapt their difficulty according to the evolving reasoning capacities of the models. To address this, we propose MorphoBench, a benchmark that incorporates multidisciplinary questions to evaluate the reasoning capabilities of large models and can adjust and update question difficulty based on the reasoning abilities of advanced models. Specifically, we curate the benchmark by selecting and collecting complex reasoning questions from existing benchmarks and sources such as Olympiad-level competitions. Additionally, MorphoBench adaptively modifies the analytical challenge of questions by leveraging key statements generated during the model's reasoning process. Furthermore, it includes questions generated using simulation software, enabling dynamic adjustment of benchmark difficulty with minimal resource consumption. We have gathered over 1,300 test questions and iteratively adjusted the difficulty of MorphoBench based on the reasoning capabilities of models such as o3 and GPT-5. MorphoBench enhances the comprehensiveness and validity of model reasoning evaluation, providing reliable guidance for improving both the reasoning abilities and scientific robustness of large models. The code has been released in https://github.com/OpenDCAI/MorphoBench.

cs.AI↗

arXiv2025

Asymptotic-preserving semi-Lagrangian discontinuous Galerkin schemes for the Boltzmann equation

In this work, we present an asymptotic-preserving semi-Lagrangian discontinuous Galerkin scheme for the Boltzmann equation that effectively handles multi-scale transport phenomena. The main challenge lies in designing appropriate moments update for penalization within the semi-Lagrangian framework. Inspired by [M. Ding, J. M. Qiu, and R. Shu, Multiscale Model. Simul. 21 (2023), no. 1, 143--167], the key ingredient is utilizing the Shu-Osher form of the scheme in the implicit-explicit Runge-Kutta (IMEX-RK) setting, which enables us to capture the correct limiting system by constructing an appropriate moments update procedure. Our theoretical analysis establishes accuracy order conditions for both the IMEX-RK time integration and the new moments update step. We also employ hypocoercivity techniques to establish stability for the linearized model. Numerical experiments for various test problems validate our proposed scheme's accuracy, asymptotic-preserving property, and robustness in various regimes, which demonstrates its effectiveness for multi-scale kinetic simulations.

math.NA↗

arXiv2025

SPOGW: a Score-based Preference Optimization method via Group-Wise comparison for workflows

Large language models (LLMs) have exhibited significant capabilities in addressing challenging problems throughout various fields, often through the use of agentic workflows that adhere to structured instructions and multi-step procedures. However, designing such workflows demands substantial manual effort, posing challenges to scalability and generalizability. Recent studies have aimed to minimize the human intervention needed for their construction, leading to advances in automated techniques for optimizing agentic workflows. However, current approaches are often constrained by their limited representational capacity, insufficient adaptability, weak scalability, and pairwise comparison paradigm -- issues that stem primarily from a dependence on discrete optimization techniques. To overcome these limitations, we introduce a new score-based preference approach, refereed as SPOGW, which operates directly on cardinal reward signals through group-wise comparison and enables more efficient and stable optimization in a continuous space. SPOGW incorporates Iterative offline GRPO (ioGRPO) with advantage-masked KL divergence (mKL), which regulates training update by placing greater emphasis on the advantageous regions of the policy response. In five benchmark datasets covering mathematical reasoning, coding, and question answering, SPOGW matches or exceeds the performance of current state-of-the-art approaches, presenting a viable and forward-looking methodology for automated generation and optimization of agentic workflows.

cs.AI↗

arXiv2025

From Long Videos to Engaging Clips: A Human-Inspired Video Editing Framework with Multimodal Narrative Understanding

The rapid growth of online video content, especially on short video platforms, has created a growing demand for efficient video editing techniques that can condense long-form videos into concise and engaging clips. Existing automatic editing methods predominantly rely on textual cues from ASR transcripts and end-to-end segment selection, often neglecting the rich visual context and leading to incoherent outputs. In this paper, we propose a human-inspired automatic video editing framework (HIVE) that leverages multimodal narrative understanding to address these limitations. Our approach incorporates character extraction, dialogue analysis, and narrative summarization through multimodal large language models, enabling a holistic understanding of the video content. To further enhance coherence, we apply scene-level segmentation and decompose the editing process into three subtasks: highlight detection, opening/ending selection, and pruning of irrelevant content. To facilitate research in this area, we introduce DramaAD, a novel benchmark dataset comprising over 800 short drama episodes and 500 professionally edited advertisement clips. Experimental results demonstrate that our framework consistently outperforms existing baselines across both general and advertisement-oriented editing tasks, significantly narrowing the quality gap between automatic and human-edited videos.

cs.CV↗

arXiv2025

ContextPRM: Leveraging Contextual Coherence for multi-domain Test-Time Scaling

Process reward models (PRMs) have demonstrated significant efficacy in enhancing the mathematical reasoning capabilities of large language models (LLMs) by leveraging test-time scaling (TTS). However, while most PRMs exhibit substantial gains in mathematical domains, the scarcity of domain-specific training data and knowledge-based learning patterns limits their generalization ability when faced with other domains. To address this limitation, we shift the learning objective from verifying domain-specific knowledge to modeling domain-agnostic logical flow. Centering on contextual coherence between chain-of-thought (CoT) steps, our approach is realized through a novel data annotation and training framework, which enhances the model's generalization capabilities across diverse domains. For instance, our resulting model, ContextPRM, achieves a notable 6.5% average accuracy improvement over the majority voting baseline via weighted majority voting across nine non-mathematical domains in MMLU-Pro, including law, history, and philosophy, significantly surpassing the 2.2% improvement from VersaPRM and 0.5% gains from other mathematics-focused PRMs, demonstrating consistent performance across both mathematical and non-mathematical domains.

cs.AI↗

arXiv2025

Human vs. AI Safety Perception? Decoding Human Safety Perception with Eye-Tracking Systems, Street View Images, and Explainable AI

The way residents perceive safety plays an important role in how they use public spaces. Studies have combined large-scale street view images and advanced computer vision techniques to measure the perception of safety of urban environments. Despite their success, such studies have often overlooked the specific environmental visual factors that draw human attention and trigger people's feelings of safety perceptions. In this study, we introduce a computational framework that enriches the existing body of literature on place perception by using eye-tracking systems with street view images and deep learning approaches. Eye-tracking systems quantify not only what users are looking at but also how long they engage with specific environmental elements. This allows us to explore the nuance of which visual environmental factors influence human safety perceptions. We conducted our research in Helsingborg, Sweden, where we recruited volunteers outfitted with eye-tracking systems. They were asked to indicate which of the two street view images appeared safer. By examining participants' focus on specific features using Mean Object Ratio in Highlighted Regions (MoRH) and Mean Object Hue (MoH), we identified key visual elements that attract human attention when perceiving safe environments. For instance, certain urban infrastructure and public space features draw more human attention while the sky is less relevant in influencing safety perceptions. These insights offer a more human-centered understanding of which urban features influence human safety perceptions. Furthermore, we compared the real human attention from eye-tracking systems with attention maps obtained from eXplainable Artificial Intelligence (XAI) results. Several XAI models were tested, and we observed that XGradCAM and EigenCAM most closely align with human safety perceptual patterns.

cs.HC↗

arXiv2025

Accelerating LLM Inference via Dynamic KV Cache Placement in Heterogeneous Memory System

Large Language Model (LLM) inference is increasingly constrained by memory bandwidth, with frequent access to the key-value (KV) cache dominating data movement. While attention sparsity reduces some memory traffic, the relevance of past tokens varies over time, requiring the full KV cache to remain accessible and sustaining pressure on both bandwidth and capacity. With advances in interconnects such as NVLink and LPDDR5X, modern AI hardware now integrates high-bandwidth memory (HBM) with high-speed off-package DRAM, making heterogeneous memory systems a practical solution. This work investigates dynamic KV cache placement across such systems to maximize aggregated bandwidth utilization under capacity constraints. Rather than proposing a specific scheduling policy, we formulate the placement problem mathematically and derive a theoretical upper bound, revealing substantial headroom for runtime optimization. To our knowledge, this is the first formal treatment of dynamic KV cache scheduling in heterogeneous memory systems for LLM inference.

cs.AR↗

arXiv2025

Breaking the HBM Bit Cost Barrier: Domain-Specific ECC for AI Inference Infrastructure

High-Bandwidth Memory (HBM) delivers exceptional bandwidth and energy efficiency for AI workloads, but its high cost per bit, driven in part by stringent on-die reliability requirements, poses a growing barrier to scalable deployment. This work explores a system-level approach to cost reduction by eliminating on-die ECC and shifting all fault management to the memory controller. We introduce a domain-specific ECC framework combining large-codeword Reed--Solomon~(RS) correction with lightweight fine-grained CRC detection, differential parity updates to mitigate write amplification, and tunable protection based on data importance. Our evaluation using LLM inference workloads shows that, even under raw HBM bit error rates up to $10^{-3}$, the system retains over 78\% of throughput and 97\% of model accuracy compared with systems equipped with ideal error-free HBM. By treating reliability as a tunable system parameter rather than a fixed hardware constraint, our design opens a new path toward low-cost, high-performance HBM deployment in AI infrastructure.

cs.AR↗

arXiv2025

Hypocoercivity for the Linear Semiconductor Boltzmann Equation with Boundaries and Uncertainties

In this paper, we establish hypocoercivity for the semiconductor Boltzmann equation with the presence of an external electrical potential under the Maxwell boundary condition. We will construct a modified entropy Lyapunov functional, which is proved to be equivalent to some weighted norm of the corresponding function space. We then show that the entropy functional dissipates along the solutions, and the exponential decay to the equilibrium state of the system follows by a Gronwall type inequality. We also generalize our arguments to situations where uncertainties in our model arise,and the hypocoercivity method we have established is adopted to analyze the regularity of the solutions along the random space.

math.AP↗

arXiv2025

A Hybrid APIM-CFGM Model for Longitudinal Non-Exchangeable Dyads: Demonstrating and Comparing Estimation Approaches Using Multilevel Modeling

Understanding change over time within dyads, such as mentor-mentee or therapist-client pairs, poses unique challenges, particularly in studies with small samples and distinguishable roles. This paper introduces a flexible hybrid longitudinal modeling that integrates features of the Actor-Partner Interdependence Model (APIM) and the Common Fate Growth Model (CFGM) to simultaneously estimate individual-level and shared dyad-level effects. Using a hypothetical peer-mentoring example (novices paired with experts), the model addresses three key issues: (1) how the interpretation of model parameters when role is dummy-coded versus effect-coded; (2) how model performance is affected by small sample sizes; and (3) how results differ between Maximum Likelihood (ML) and Bayesian estimation. Simulated data for 50 dyads across five time points are analyzed, with subsampling at 30 and 5 dyads. Models are estimated in a multilevel modeling framework using R (lme4 for ML and brms for Bayesian inference). Results show that dummy and effect coding reflect distinct interpretations: dummy coding expresses effects relative to a reference group, whereas effect coding centers parameters on the grand mean across roles. Although model fit remains unchanged, the choice of coding impacts how group-level effects and interactions are interpreted. Very small samples (e.g., 5 dyads) lead to unstable estimates, whereas Bayesian credible intervals more accurately reflect uncertainty in such cases. In larger samples, ML and Bayesian estimates converge. This APIM-CFGM hybrid offers a practical tool for researchers analyzing longitudinal dyadic data with distinguishable roles and smaller sample sizes.

stat.ME↗