arXiv subjects

Lu Chen

Publications and source records attributed to Lu Chen.

At least 127 records · Page 7Linked to original sources

arXiv2025

Enhancing LLM Reliability via Explicit Knowledge Boundary Modeling

Large language models (LLMs) are prone to hallucination stemming from misaligned self-awareness, particularly when processing queries exceeding their knowledge boundaries. While existing mitigation strategies employ uncertainty estimation or query rejection mechanisms, they suffer from computational efficiency and sacrificed helpfulness. To address these issues, we propose the Explicit Knowledge Boundary Modeling (EKBM) framework, integrating fast and slow reasoning systems to harmonize reliability and usability. The framework first employs a fast-thinking model to generate confidence-labeled responses, enabling immediate utilization of high-confidence outputs, whereas uncertain predictions trigger a slow refinement model for accuracy improvement. To align model behavior with our proposed object, we propose a hybrid training pipeline, enhancing self-awareness without degrading task performance. Evaluations on dialogue state tracking tasks demonstrate that EKBM achieves superior model reliability over uncertainty-based baselines. Further analysis reveals that refinement substantially boosts accuracy while maintaining low computational overhead. The framework establishes a scalable paradigm for deploying reliable LLMs in error-sensitive applications, effectively balancing accuracy and practical utility.

cs.CL↗

arXiv2025

When Long Helps Short: How Context Length in Supervised Fine-tuning Affects Behavior of Large Language Models

Large language models (LLMs) have achieved impressive performance across natural language processing (NLP) tasks. As real-world applications increasingly demand longer context windows, continued pretraining and supervised fine-tuning (SFT) on long-context data has become a common approach. While the effects of data length in continued pretraining have been extensively studied, their implications for SFT remain unclear. In this work, we systematically investigate how SFT data length influences LLM behavior on short-context tasks. Counterintuitively, we find that long-context SFT improves short-context performance, contrary to the commonly observed degradation from long-context pretraining. To uncover the underlying mechanisms of this phenomenon, we first decouple and analyze two key components, Multi-Head Attention (MHA) and Feed-Forward Network (FFN), and show that both independently benefit from long-context SFT. We further study their interaction and reveal a knowledge preference bias: long-context SFT promotes contextual knowledge, while short-context SFT favors parametric knowledge, making exclusive reliance on long-context SFT suboptimal. Finally, we demonstrate that hybrid training mitigates this bias, offering explainable guidance for fine-tuning LLMs.

cs.CL↗

arXiv2025

Moon: A Modality Conversion-based Efficient Multivariate Time Series Anomaly Detection

Multivariate time series (MTS) anomaly detection identifies abnormal patterns where each timestamp contains multiple variables. Existing MTS anomaly detection methods fall into three categories: reconstruction-based, prediction-based, and classifier-based methods. However, these methods face two key challenges: (1) Unsupervised learning methods, such as reconstruction-based and prediction-based methods, rely on error thresholds, which can lead to inaccuracies; (2) Semi-supervised methods mainly model normal data and often underuse anomaly labels, limiting detection of subtle anomalies;(3) Supervised learning methods, such as classifier-based approaches, often fail to capture local relationships, incur high computational costs, and are constrained by the scarcity of labeled data. To address these limitations, we propose Moon, a supervised modality conversion-based multivariate time series anomaly detection framework. Moon enhances the efficiency and accuracy of anomaly detection while providing detailed anomaly analysis reports. First, Moon introduces a novel multivariate Markov Transition Field (MV-MTF) technique to convert numeric time series data into image representations, capturing relationships across variables and timestamps. Since numeric data retains unique patterns that cannot be fully captured by image conversion alone, Moon employs a Multimodal-CNN to integrate numeric and image data through a feature fusion model with parameter sharing, enhancing training efficiency. Finally, a SHAP-based anomaly explainer identifies key variables contributing to anomalies, improving interpretability. Extensive experiments on six real-world MTS datasets demonstrate that Moon outperforms six state-of-the-art methods by up to 93% in efficiency, 4% in accuracy and, 10.8% in interpretation performance.

cs.LG↗

arXiv2025

MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing

We introduce MinerU2.5, a 1.2B-parameter document parsing vision-language model that achieves state-of-the-art recognition accuracy while maintaining exceptional computational efficiency. Our approach employs a coarse-to-fine, two-stage parsing strategy that decouples global layout analysis from local content recognition. In the first stage, the model performs efficient layout analysis on downsampled images to identify structural elements, circumventing the computational overhead of processing high-resolution inputs. In the second stage, guided by the global layout, it performs targeted content recognition on native-resolution crops extracted from the original image, preserving fine-grained details in dense text, complex formulas, and tables. To support this strategy, we developed a comprehensive data engine that generates diverse, large-scale training corpora for both pretraining and fine-tuning. Ultimately, MinerU2.5 demonstrates strong document parsing ability, achieving state-of-the-art performance on multiple benchmarks, surpassing both general-purpose and domain-specific models across various recognition tasks, while maintaining significantly lower computational overhead.

cs.CV↗

arXiv2025

Early Prediction of In-Hospital ICU Mortality Using Innovative First-Day Data: A Review

The intensive care unit (ICU) manages critically ill patients, many of whom face a high risk of mortality. Early and accurate prediction of in-hospital mortality within the first 24 hours of ICU admission is crucial for timely clinical interventions, resource optimization, and improved patient outcomes. Traditional scoring systems, while useful, often have limitations in predictive accuracy and adaptability. Objective: This review aims to systematically evaluate and benchmark innovative methodologies that leverage data available within the first day of ICU admission for predicting in-hospital mortality. We focus on advancements in machine learning, novel biomarker applications, and the integration of diverse data types.

cs.LG↗

arXiv2025

EgoAgent: A Joint Predictive Agent Model in Egocentric Worlds

Learning an agent model that behaves like humans-capable of jointly perceiving the environment, predicting the future, and taking actions from a first-person perspective-is a fundamental challenge in computer vision. Existing methods typically train separate models for these abilities, which fail to capture their intrinsic relationships and prevent them from learning from each other. Inspired by how humans learn through the perception-action loop, we propose EgoAgent, a unified agent model that simultaneously learns to represent, predict, and act within a single transformer. EgoAgent explicitly models the causal and temporal dependencies among these abilities by formulating the task as an interleaved sequence of states and actions. It further introduces a joint embedding-action-prediction architecture with temporally asymmetric predictor and observer branches, enabling synergistic optimization across all three capabilities. Comprehensive evaluations of EgoAgent on representative tasks such as image classification, egocentric future state prediction, and 3D human motion prediction demonstrate the superiority of our method. The code and trained models will be publicly available at https://github.com/zju3dv/EgoAgent.

cs.CV↗

arXiv2025

Efficient Size Constraint Community Search over Heterogeneous Information Networks

The goal of community search in heterogeneous information networks (HINs) is to identify a set of closely related target nodes that includes a query target node. In practice, a size constraint is often imposed due to limited resources, which has been overlooked by most existing HIN community search works. In this paper, we introduce the size-bounded community search problem to HIN data. Specifically, we propose a refined (k, P)-truss model to measure community cohesiveness, aiming to identify the most cohesive community of size s that contains the query node. We prove that this problem is NP-hard. To solve this problem, we develop a novel B\&B framework that efficiently generates target node sets of size s. We then tailor novel bounding, branching, total ordering, and candidate reduction optimisations, which enable the framework to efficiently lead to an optimum result. We also design a heuristic algorithm leveraging structural properties of HINs to efficiently obtain a high-quality initial solution, which serves as a global lower bound to further enhance the above optimisations. Building upon these, we propose two exact algorithms that enumerate combinations of edges and nodes, respectively. Extensive experiments on real-world datasets demonstrate the effectiveness and efficiency of the proposed methods.

cs.DB↗

arXiv2025

Electronic mean free path of the cuprate superconductor Bi$_2$Sr$_2$CaCu$_2$O$_{8+δ}$ from thermal Hall conductivity

We use thermal transport to access the electronic mean free path of $d$-wave quasiparticles in one of the most widely studied cuprate superconductors, Bi$_2$Sr$_2$CaCu$_2$O$_{8+δ}$ (Bi2212). We have measured the thermal conductivity $κ_{\rm xx}$ and the thermal Hall conductivity $κ_{\rm xy}$ of three single crystals across a range of dopings. In the overdoped and optimally-doped samples, a clear enhancement is observed in both $κ_{\rm xx}$ and $κ_{\rm xy}$ upon cooling below the critical temperature $T_{\rm c}$, due to a suppression of the inelastic electron-electron scattering as electrons condense into pairs. The underdoped sample shows no enhancement in either, pointing to a high degree of disorder in that sample. For the two highest dopings, the magnitude of the enhancement in $κ_{\rm xy}$ is controlled by the strength of the elastic impurity scattering. Using a prior model to estimate the mean free path from $κ_{\rm xy}$ data, we find that the mean free path in Bi2212 is approximately 7 times shorter than in YBa$_2$Cu$_3$O$_7$, considered to be one of the least disordered cuprates. We conclude that the thermal Hall technique is a good way to compare the mean free path of $d$-wave quasiparticles in various cuprate materials.

cond-mat.supr-con↗

arXiv2025

Compressing KV Cache for Long-Context LLM Inference with Inter-Layer Attention Similarity

The rapid expansion of context window sizes in Large Language Models~(LLMs) has enabled them to tackle increasingly complex tasks involving lengthy documents. However, this progress comes at the cost of a substantial increase in memory usage during inference, primarily due to the linear growth of the key-value~(KV) cache. Existing KV cache compression methods often discard less relevant tokens, which can lead to significant performance degradation when critical information is lost. In this paper, we propose \textsc{PoD}~(Proximal tokens over Distant tokens), a novel KV cache compression framework that allocates memory according to token importance, retaining less important tokens in a more compact, shared form rather than discarding them entirely. Our approach is motivated by two key observations: (1) proximal tokens -- those at the beginning and end of the context -- are significantly more important for next-token prediction, and (2) attention scores for distant tokens are highly redundant across consecutive layers. Leveraging these insights, \textsc{PoD} preserves the full KV cache for proximal tokens, while for distant tokens, it shares key states across layers. Since attention scores are determined by both queries and keys, sharing key states enables multiple layers to reuse a single set of keys for distant tokens, substantially reducing KV cache memory without discarding essential context. We further introduce a lightweight post-training adaptation to enable the model to adjust to this new attention-sharing structure. Extensive experiments on both synthetic~(Needle in a Haystack) and real-world long-context benchmarks demonstrate that \textsc{PoD} reduces KV cache memory usage by up to 35\% without compromising performance. Our method is orthogonal to existing token-selection-based techniques and can be combined with them for further KV cache compression.

cs.CL↗

arXiv2025

GSIM: Accelerating RTL Simulation for Large-Scale Designs

Register Transfer Level (RTL) simulation is widely used in design space exploration, verification, debugging, and preliminary performance evaluation for hardware design. Among various RTL simulation approaches, software simulation is the most commonly used due to its flexibility, low cost, and ease of debugging. However, the slow simulation of complex designs has become the bottleneck in design flow. In this work, we explore the sources of computation overhead of RTL simulation and conclude them into four factors. To optimize these factors, we propose several techniques at the supernode level, node level, and bit level. Finally, we implement these techniques in a novel RTL simulator GSIM. GSIM succeeds in simulating XiangShan, the state-of-the-art open-source RISC-V processor. Besides, compared to Verilator, GSIM can achieve speedup of 7.34x for booting Linux on XiangShan, and 19.94x for running CoreMark on Rocket.

cs.AR↗

arXiv2025

OS-MAP: How Far Can Computer-Using Agents Go in Breadth and Depth?

Computer-using agents have shown strong potential to boost human productivity and enable new application forms across platforms. While recent advances have led to usable applications, existing benchmarks fail to account for the internal task heterogeneity and the corresponding agent capabilities, as well as their alignment with actual user demands-hindering both targeted capability development and the reliable transition of research progress into practical deployment. To bridge the gap, we present OS-MAP, a benchmark for daily computer-using automation that organizes its 416 realistic tasks across 15 applications along two key dimensions: a five-level taxonomy of automation and a generalization scope derived from a real-world user demand hierarchy. To enable fine-grained analysis of required capabilities and alignment with real-world scenarios, OS-MAP evaluates agents along two dimensions: automation level across a five-level taxonomy, and generalization scope across a demand hierarchy. This design captures varying levels of required agent autonomy and generalization, forming a performance-generalization evaluation matrix for structured and comprehensive assessment. Experiments show that even State-of-the-Art agents with VLM backbones struggle with higher-level tasks involving perception, reasoning, and coordination-highlighting the need for a deeper understanding of current strengths and limitations to drive the future progress in computer-using agents research and deployment. All code, environments, baselines, and data are publicly available at https://github.com/OS-Copilot/OS-Map.

cs.AI↗

arXiv2025

Asymptotically sharp stability of Sobolev inequalities on the Heisenberg group with dimension-dependent constants

In this paper, we are concerned with the optimal asymptotic lower bound for the stability of Sobolev inequality on the Heisenberg group. We first establish the optimal local stability of Sobolev inequality on the CR sphere through bispherical harmonics and complicated orthogonality technique ( see Lemma 3.1). The loss of rearrangement inequality in the CR setting makes it impossible to use any rearrangement flow technique (either differential rearrangement flow or integral rearrangement flow) to derive the optimal stability of Sobolev inequality on the CR sphere from corresponding optimal local stability. To circumvent this, we will use the CR Yamabe flow to establish the optimal stability of Sobolev inequality on the Heisenberg group with the dimension-dependent constants (see Theorem 1.1). As an application, we also establish the optimal stability of the Hardy-Littlewood-Sobolev (HLS) inequality for special conformal index with the dimension-dependent constants (see Theorem 1.3). Our approach is rearrangement-free and can be used to study the optimal stability problem for fractional Sobolev inequality or HLS inequality on the Heisenberg group once the corresponding continuous flow is established.

math.AP↗

arXiv2025

Light-Induced Giant Enhancement of the Nonlinear Hall Effect in Two-Dimensional Electron Gases at KTaO3 (111) Interfaces

The nonlinear Hall effect (NLHE), an emergent phenomenon in noncentrosymmetric systems, enables the generation of a transverse voltage without an external magnetic field through a second-order electrical response. However, achieving a sizable NLHE signal remains a critical challenge for its application in frequency-doubling and rectifying devices. Here, we report a light-induced giant enhancement of the NLHE in the two-dimensional electron gas (2DEG) at the CaZrO3/KTaO3 (111) interface. Under light illumination, the second harmonic Hall voltage (V2ω y) increases substantially and undergoes a sign reversal. Correspondingly,the second-order transverse conductivity increases by nearly five orders of magnitude, reaching 2.4 um V-1 omega-1, while also reversing its sign. Scaling analysis indicates that skew scattering is the dominant mechanism underlying the NLHE and is highly tunable via optical gating. Photoexcitation pumps electrons from in-gap states into the higher-lying Ta 5d conduction band, generating high-mobility photocarriers that significantly increase the cubic transport scattering time, thereby driving a dramatic enhancement of σ(2) yxx. First-principles calculations further reveal that the Berry curvature distribution on the Fermi surface strongly depends on band filling. As the Fermi level approaches a band crossing in the Ta 5d subband near the M point, the Berry curvature triple undergoes a sign change, accounting for the experimentally observed sign reversal of the nonlinear Hall response. Our work offers a new strategy to optically boost and tune the nonlinear Hall effect in oxide 2DEG systems, paving the way for applications in light-controlled rectification and nonlinear electronic devices.

cond-mat.mes-hall↗

arXiv2025

NavComposer: Composing Language Instructions for Navigation Trajectories through Action-Scene-Object Modularization

Language-guided navigation is a cornerstone of embodied AI, enabling agents to interpret language instructions and navigate complex environments. However, expert-provided instructions are limited in quantity, while synthesized annotations often lack quality, making them insufficient for large-scale research. To address this, we propose NavComposer, a novel framework for automatically generating high-quality navigation instructions. NavComposer explicitly decomposes semantic entities such as actions, scenes, and objects, and recomposes them into natural language instructions. Its modular architecture allows flexible integration of state-of-the-art techniques, while the explicit use of semantic entities enhances both the richness and accuracy of instructions. Moreover, it operates in a data-agnostic manner, supporting adaptation to diverse navigation trajectories without domain-specific training. Complementing NavComposer, we introduce NavInstrCritic, a comprehensive annotation-free evaluation system that assesses navigation instructions on three dimensions: contrastive matching, semantic consistency, and linguistic diversity. NavInstrCritic provides a holistic evaluation of instruction quality, addressing limitations of traditional metrics that rely heavily on expert annotations. By decoupling instruction generation and evaluation from specific navigation agents, our method enables more scalable and generalizable research. Extensive experiments provide direct and practical evidence for the effectiveness of our method.

cs.AI↗

arXiv2025

OneDB: A Distributed Multi-Metric Data Similarity Search System

Increasingly massive volumes of multi-modal data are being accumulated in many {real world} settings, including in health care and e-commerce. This development calls for effective general-purpose data management solutions for multi-modal data. Such a solution must facilitate user-friendly and accurate retrieval of any multi-modal data according to diverse application requirements. Further, such a solution must be capable of efficient and scalable retrieval. To address this need, we present OneDB, a distributed multi-metric data similarity retrieval system. This system exploits the fact that data of diverse modalities, such as text, images, and video, can be represented as metric data. The system thus affords each data modality its own metric space with its own distance function and then uses a multi-metric model to unify multi-modal data. The system features several innovations: (i) an extended Spart SQL query interface; (ii) lightweight means of learning appropriate weights of different modalities when retrieving multi-modal data to enable accurate retrieval; (iii) smart search-space pruning strategies that improve efficiency; (iv) two-layered indexing of data to ensure load-balancing during distributed processing; and (v) end-to-end system parameter autotuning. Experiments on three real-life datasets and two synthetic datasets offer evidence that the system is capable of state-of-the-art performance: (i) efficient and effective weight learning; (ii) retrieval accuracy improvements of 12.63\%--30.75\% over the state-of-the-art vector similarity search system at comparable efficiency; (iii) accelerated search by 2.5--5.75x over state-of-the-art single- or multi-metric solutions; (iv) demonstrated high scalability; and (v) parameter tuning that enables performance improvements of 15+%.

cs.DB↗

arXiv2025

Developing ChemDFM as a large language foundation model for chemistry

Artificial intelligence (AI) has played an increasingly important role in chemical research. However, most models currently used in chemistry are specialist models that require training and tuning for specific tasks. A more generic and efficient solution would be an AI model that could address many tasks and support free-form dialogue in the broad field of chemistry. In its utmost form, such a generalist AI chemist could be referred to as Chemical General Intelligence. Large language models (LLMs) have recently logged tremendous success in the general domain of natural language processing, showing emerging task generalization and free-form dialogue capabilities. However, domain knowledge of chemistry is largely missing when training general-domain LLMs. The lack of such knowledge greatly hinders the performance of generalist LLMs in the field of chemistry. To this end, we develop ChemDFM, a pioneering LLM for chemistry trained on 34B tokens from chemical literature and textbooks, and fine-tuned using 2.7M instructions. As a result, it can understand and reason with chemical knowledge in free-form dialogue. Quantitative evaluations show that ChemDFM significantly surpasses most representative open-source LLMs. It outperforms GPT-4 on a great portion of chemical tasks, despite the substantial size difference. We have open-sourced the inference codes, evaluation datasets, and model weights of ChemDFM on Huggingface (https://huggingface.co/OpenDFM/ChemDFM-v1.0-13B).

cs.CL↗

arXiv2025

Existence and Nonexistence of Extremals for Trudinger-Moser inequalities with $L^p$ type perturbation on any bounded planar domains

In this study, we investigate the perturbed Trudinger-Moser inequalities as follows:\[ S_Ω(λ,p)=\sup_{u\in H_{0}^{1}(Ω),\Vert\nabla u\Vert _{L^{2}\left( Ω\right) }\leq 1}\int_Ω\left( e^{4πu^{2}}-λ|u|^{p}\right) dx, \] where $1\leq p<\infty$ and $Ω$ is a bounded domain in $\mathbb{R}^2$. Our results demonstrate that there exists a threshold $λ^{\ast}(p)>0$ such that $S_Ω(λ,p)$ is attainable if $λ<λ^{\ast}(p)$, but unattainable if $λ>λ^{\ast}(p)$ when $p\in[1,2]$. For $p>2$, however, we show that $S_Ω(λ,p)$ is always attainable for any $λ\in \mathbb{R}$. These results are achieved through a refined blow-up analysis, which allow us to establish a sharp Dirichlet energy expansion formula for sequences of solutions to the corresponding Euler-Lagrange equations. The asymmetric nature of our problem poses significant challenges to our analysis. To address these, we will establish an appropriate comparison principle between radial and non-radial solutions of the associated Euler-Lagrange equations. Our study establishes a complete characterization of how $L^p$-type perturbations influence the existence of extremals for critical Trudinger-Moser inequalities on any bounded planar domains, this extends the classical Brezis-Nirenberg problem framework to the two-dimensional settings.

math.AP↗

arXiv2025

Few-Shot Identity Adaptation for 3D Talking Heads via Global Gaussian Field

Reconstruction and rendering-based talking head synthesis methods achieve high-quality results with strong identity preservation but are limited by their dependence on identity-specific models. Each new identity requires training from scratch, incurring high computational costs and reduced scalability compared to generative model-based approaches. To overcome this limitation, we propose FIAG, a novel 3D speaking head synthesis framework that enables efficient identity-specific adaptation using only a few training footage. FIAG incorporates Global Gaussian Field, which supports the representation of multiple identities within a shared field, and Universal Motion Field, which captures the common motion dynamics across diverse identities. Benefiting from the shared facial structure information encoded in the Global Gaussian Field and the general motion priors learned in the motion field, our framework enables rapid adaptation from canonical identity representations to specific ones with minimal data. Extensive comparative and ablation experiments demonstrate that our method outperforms existing state-of-the-art approaches, validating both the effectiveness and generalizability of the proposed framework. Code is available at: \textit{https://github.com/gme-hong/FIAG}.

cs.CV↗