arXiv subjects

Chuan Chen

Publications and source records attributed to Chuan Chen.

At least 19 recordsLinked to original sources

arXiv2026

Structural Entropy-Driven Graph Diffusion Generation for One-Shot Federated Graph Learning

One-shot federated graph learning (FGL) requires the server to estimate client contributions from highly compressed information, yet conventional volume-based weighting captures the amount of client data while overlooking how its connectivity is organized. In this paper, we propose SPIRE, a Structural Entropy-Driven Graph Diffusion Generation method that introduces topology-aware client differentiation into one-shot FGL. Specifically, we employ first-order degree-distribution structural entropy as a compact descriptor of degree-mass dispersion and use it to derive structural client weights, providing an inductive bias that accounts for differences in graph topology beyond data volume. On the generation side, a graph diffusion model on the server synthesizes pseudographs conditioned on the weighted client prototypes, capturing both semantic and structural information without requiring additional client-side training. The generated pseudographs are then assembled via disjoint union fusion to train a global graph neural network. Extensive experiments on seven real-world graph datasets demonstrate that SPIRE consistently outperforms conventional and one-shot FGL methods, with particularly strong gains under highly heterogeneous (non-IID) and graph-perturbed settings.

cs.LG↗

arXiv2026

Automatic Echocardiography Segmentation via Transition Probability Correlation for Stable Semantic Extraction

While echocardiography is essential for cardiovascular diagnosis, inherent speckle noise and low signal-to-noise ratio often lead to ambiguous semantic features and fragmented boundaries. These limitations significantly hinder the segmentation accuracy of deep learning models in complex clinical cases. Moreover, temporal motion of the heart plays a critical role in recognizing anatomical structures. To address these challenges, we designed a STLSF module which comprises a window-matching-based semantic correction component and a semantics-guided texture enhancement component. By leveraging local transition probability correlations to correct semantics and employing semantics-guided texture enhancement, the STLSF module effectively mitigates texture instability and ambiguous semantic interpretations caused by disadvantaged echocardiography quality. Additionally, to facilitate the encoder's adaptation to the intrinsic priors of ultrasound-specific imaging patterns, we propose a frequency-aware denoising pre-training method. The entire work builds a convolution-based network with locality inductive bias and long-range dependencies. Extensive experiments confirm our SOTA performance, achieving 93.87\% Dice on CAMUS and 92.62\% on EchoNet-Dynamic, with respective HD95 values of 3.29mm and 2.73mm.

cs.CV↗

arXiv2026

Evidence for Deconfined Magnetic Order in the Kitaev-$J_3$ Model

We investigate the Kitaev-$J_3$ honeycomb model using variational Monte Carlo calculations combined with a vison-quasiparticle analysis of the parent Kitaev spin liquid (KSL). We provide evidence for deconfined magnetic phases in which zigzag or antiferromagnetic order coexists with remnant $\mathbb{Z}_2$ topological structure inherited from the KSL. The optimized variational wave functions retain multiple linearly independent topological sectors on a torus, whereas those of conventional ordered phases collapse to a single sector. The vison-quasiparticle analysis shows that magnetic order naturally arises from vison-pair condensation while single visons remain gapped, yielding a microscopic mechanism for magnetic ordering without immediate confinement. The resulting phases further host gapless spinons with multiple Majorana cones, offering a possible microscopic scenario for the anomalous low-temperature longitudinal thermal transport reported in magnetically ordered Kitaev materials such as Na$_2$Co$_2$TeO$_6$. Our results reveal a microscopic route to fractionalized magnetism beyond the conventional dichotomy between magnetic order and spin-liquid behavior.

cond-mat.str-el↗

arXiv2026

LayersReg: A Layer-by-Layer Progressive Regressor for Reliable Intraoperative 3D/2D Registration

3D/2D registration serves as a cornerstone technique in surgical navigation. Traditional iterative optimization algorithms suffer from low efficiency and high failure rates in intraoperative settings. Deep learning-based methods reformulate registration from iterative optimization to a regression problem that maps image appearance features to spatial pose, typically achieving improved real-time performance and accuracy. However, such learnable methods are confined to memory-driven retrieval of specific pose features rather than understanding the task of image alignment itself, which limits their generalization in complex scenarios. We propose LayersReg, a pioneering regression paradigm that endows the model with 3D anatomical awareness and searches for the correct pose in a progressive, layer-by-layer manner. Inspired by the iterative pose-searching optimization criterion of classical registration, LayersReg searches for correlations between the moving and fixed images in feature space, capturing the trend of pixel flow and thereby converging iteratively toward the correct spatial pose transformation. We further design a coupling of node-wise regression with the progressive registration framework to enhance the model's perception of spatial pose changes. Experimental results demonstrate that under large offsets and multimodality conditions, LayersReg achieves high accuracy on both X-ray/CT registration (0.68{\deg}, 1.41 mm) and slice localization (0.73{\deg}, 1.55 mm) tasks, outperforming existing state-of-the-art methods while meeting the intraoperative demands for precision and real-time capability.

cs.CV↗

arXiv2026

DataClawBench: An Agent Benchmark for Exploratory Real-World Financial Data Analysis

Autonomous data analysis agents are increasingly expected to conduct exploratory analysis with limited human guidance about data. However, existing benchmarks typically evaluate such agents in prior-guided settings, providing selected data sources, explicit data schemas, or cleaned data, thereby understating the exploratory burden. To evaluate this realistic exploratory data analysis task, we introduce DataClawBench, a benchmark built from financial think-tank consulting scenarios where agents must independently explore unfamiliar, noisy, cross-domain data and produce verifiable conclusions. DataClawBench provides a unified real-world data environment with approximately 2.06 million records across enterprise, industry, and policy domains, with native data noise preserved. On top of this data environment, it defines 492 multi-step cross-domain tasks, each annotated with intermediate milestones that diagnose exploration and reasoning failures beyond outcome accuracy. A systematic evaluation of eight advanced LLMs under the OpenClaw agent reveals that exploratory data analysis breaks agent reliability: more exploration does not reliably translate into task-relevant progress or correct final answers.

cs.AI↗

arXiv2026

NTIRE 2026 The Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results

This paper presents an overview of the NTIRE 2026 Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images. Building upon the success of the first edition, this challenge attracted a wide range of impressive solutions, all developed and evaluated on our real-world Raindrop Clarity dataset~\cite{jin2024raindrop}. For this edition, we adjust the dataset with 14,139 images for training, 407 images for validation, and 593 images for testing. The primary goal of this challenge is to establish a strong and practical benchmark for the removal of raindrops under various illumination and focus conditions. In total, 168 teams have registered for the competition, and 17 teams submitted valid final solutions and fact sheets for the testing phase. The submitted methods achieved strong performance on the Raindrop Clarity dataset, demonstrating the growing progress in this challenging task.

cs.CV↗

arXiv2026

Dynamical Response of the Kitaev Spin Liquid under Third-Nearest-Neighbor Heisenberg Interaction

Motivated by growing evidence for the significance of the third-nearest-neighbor Heisenberg ($J_3$) interaction in candidate Kitaev materials, we investigate the dynamical properties of the Kitaev spin liquid (KSL) under a $J_3$ perturbation, focusing on its spin dynamical structure factor (DSF) and Raman scattering. Within a self-consistent parton mean-field plus random-phase approximation framework, we find that $J_3$ induces coherent, paramagnon-like collective modes that coexist with a high-energy Majorana continuum in the spin DSF. The softening of these modes with increasing $|J_3|$ signals a quantum phase transition to magnetic order. Remarkably, magnetic ordering sets in at a common critical $J_3$ for both ferromagnetic ($K<0$) and antiferromagnetic ($K>0$) Kitaev models, with the resulting ordered states forming exact dual pairs under a four-sublattice duality transformation that maps $(K,J_3) \rightarrow (-K,J_3)$. An external magnetic field further softens the preexisting paramagnon modes, thereby enhancing magnetic order. Perturbative Raman calculations show that while the Kitaev-like Raman vertex probes only itinerant matter Majorana fermions, the response from the $J_3$-like vertex features both matter Majoranas and visons. Four-vison excitations produce a sharp peak accompanied by a two-fermion continuum, whereas two-vison excitations yield a continuum closely resembling the single-matter-fermion density of states. These results provide a unified perspective on the dynamical signatures of $J_3$-perturbed KSL and are helpful for interpreting experimental spectra in candidate Kitaev materials with sizable $J_3$ interactions.

cond-mat.str-el↗

arXiv2026

SWE-CI: Evaluating Agent Capabilities in Maintaining Codebases via Continuous Integration

Large language model (LLM)-powered agents have demonstrated strong capabilities in automating software engineering tasks such as static bug fixing. However, in the real world, the development of mature software is typically predicated on complex requirement changes and long-term feature iterations -- a process that static, one-shot repair paradigms fail to capture. To bridge this gap, we propose SWE-CI, the first repository-level benchmark built upon the Continuous Integration loop, aiming to shift the evaluation paradigm for code generation from static, short-term functional correctness toward dynamic, long-term maintainability. The key insight is simple: Maintainability can be revealed by tracking how functional correctness changes over time. The benchmark comprises 100 tasks, each deriving from a real-world code repository with a development history spanning an average of 233 days and 71 consecutive commits. SWE-CI requires agents to systematically resolve these tasks through dozens of rounds of analysis and coding iterations. SWE-CI provides valuable insights into how well agents can sustain code quality throughout long-term evolution.

cs.SE↗

arXiv2026

AIConfigurator: Lightning-Fast Configuration Optimization for Multi-Framework LLM Serving

Optimizing Large Language Model (LLM) inference in production systems is increasingly difficult due to dynamic workloads, stringent latency/throughput targets, and a rapidly expanding configuration space. This complexity spans not only distributed parallelism strategies (tensor/pipeline/expert) but also intricate framework-specific runtime parameters such as those concerning the enablement of CUDA graphs, available KV-cache memory fractions, and maximum token capacity, which drastically impact performance. The diversity of modern inference frameworks (e.g., TRT-LLM, vLLM, SGLang), each employing distinct kernels and execution policies, makes manual tuning both framework-specific and computationally prohibitive. We present AIConfigurator, a unified performance-modeling system that enables rapid, framework-agnostic inference configuration search without requiring GPU-based profiling. AIConfigurator combines (1) a methodology that decomposes inference into analytically modelable primitives - GEMM, attention, communication, and memory operations while capturing framework-specific scheduling dynamics; (2) a calibrated kernel-level performance database for these primitives across a wide range of hardware platforms and popular open-weights models (GPT-OSS, Qwen, DeepSeek, LLama, Mistral); and (3) an abstraction layer that automatically resolves optimal launch parameters for the target backend, seamlessly integrating into production-grade orchestration systems. Evaluation on production LLM serving workloads demonstrates that AIConfigurator identifies superior serving configurations that improve performance by up to 40% for dense models (e.g., Qwen3-32B) and 50% for MoE architectures (e.g., DeepSeek-V3), while completing searches within 30 seconds on average. Enabling the rapid exploration of vast design spaces - from cluster topology down to engine specific flags.

cs.LG↗

arXiv2025

ParaAegis: Parallel Protection for Flexible Privacy-preserved Federated Learning

Federated learning (FL) faces a critical dilemma: existing protection mechanisms like differential privacy (DP) and homomorphic encryption (HE) enforce a rigid trade-off, forcing a choice between model utility and computational efficiency. This lack of flexibility hinders the practical implementation. To address this, we introduce ParaAegis, a parallel protection framework designed to give practitioners flexible control over the privacy-utility-efficiency balance. Our core innovation is a strategic model partitioning scheme. By applying lightweight DP to the less critical, low norm portion of the model while protecting the remainder with HE, we create a tunable system. A distributed voting mechanism ensures consensus on this partitioning. Theoretical analysis confirms the adjustments between efficiency and utility with the same privacy. Crucially, the experimental results demonstrate that by adjusting the hyperparameters, our method enables flexible prioritization between model accuracy and training time.

cs.LG↗

arXiv2025

Composable Score-based Graph Diffusion Model for Multi-Conditional Molecular Generation

Controllable molecular graph generation is essential for material and drug discovery, where generated molecules must satisfy diverse property constraints. While recent advances in graph diffusion models have improved generation quality, their effectiveness in multi-conditional settings remains limited due to reliance on joint conditioning or continuous relaxations that compromise fidelity. To address these limitations, we propose Composable Score-based Graph Diffusion model (CSGD), the first model that extends score matching to discrete graphs via concrete scores, enabling flexible and principled manipulation of conditional guidance. Building on this foundation, we introduce two score-based techniques: Composable Guidance (CoG), which allows fine-grained control over arbitrary subsets of conditions during sampling, and Probability Calibration (PC), which adjusts estimated transition probabilities to mitigate train-test mismatches. Empirical results on four molecular datasets show that CSGD achieves state-of-the-art performance, with a 15.3% average improvement in controllability over prior methods, while maintaining high validity and distributional fidelity. Our findings highlight the practical advantages of score-based modeling for discrete graph generation and its capacity for flexible, multi-property molecular design.

cs.LG↗

arXiv2025

Anyon polarons as a window into the competing phases of the Kitaev-Gamma-Gamma' model

We investigate the dispersions of anyon quasi-particles in the Kitaev honeycomb spin-liquid perturbed by $\Gamma$ and $\Gamma'$ couplings in order to understand phase transitions into competing states through anyon gap-closing instabilities. We demonstrate how anyon gap closings allow to understand phase transitions into a plethora of previously identified competing phases -- including zigzag, stripy, $120^\circ$, and incommensurate spiral phases -- and are in agreement with numerical studies not only on the nature of the phases, but also on the specific critical values of $\Gamma$ and $\Gamma'$ couplings. Remarkably, when the anti-ferromagnetic Kitaev model is perturbed by an ferromagnetic $\Gamma$ interaction, we find that the single-vison and fermion gaps remain open while the gap of a magnon-like local boson vanishes, implying that the resulting state has coexistence of a spontaneous broken symmetry and the fractionalization pattern of the Kitaev spin liquid. The magnetic long-range order could be either a stripy antiferromagnet or an incommensurate spiral, depending on the sign of $\Gamma'$.

cond-mat.str-el↗

arXiv2025

Reframing Spatial Dependence as Geographic Feature Attribution

Spatial dependence, referring to the correlation between variable values observed at different geographic locations, is one of the most fundamental characteristics of spatial data. The presence of spatial dependence violates the classical statistical assumption of independent and identically distributed observations and implies a high degree of information redundancy within spatial datasets. However, this redundancy can also be interpreted as structured information, which has been widely leveraged in spatial modeling, prediction, and explanation tasks. With the rise of geospatial big data and the rapid advancement of deep learning and large models, effectively modeling and characterizing spatial dependence has become essential for enhancing the performance of spatial analysis and uncovering latent spatial processes. From a data-driven perspective, this study proposes a novel interpretation: spatial dependence can be understood as the contribution of geographic location -- specifically, latitude and longitude -- to the observed variation in target variables. To validate this hypothesis, we conduct simulation experiments in which data are generated based on known spatial processes. We train machine learning models to predict variable values using only coordinate information. Subsequently, XAI techniques are employed to quantify the contribution of spatial features. The resulting importance scores are then compared with local indicators of spatial association (LISA). Across a range of spatial process settings, we observe consistently high correlations (greater than 0.94) between coordinate-based contributions and LISA values. These findings offer a new data-driven perspective on spatial dependence, bridging traditional spatial statistical approaches with modern machine learning techniques.

physics.soc-ph↗

arXiv2025

Getting There and Getting In: How Mobility and Sorting Keep Women out of Top Startup Accelerators

Startup accelerators are a leading gateway to venture capital, but top programs often require founders to relocate to a venture hub. From a hand-collected census of U.S. accelerator startups (2008-2011) followed for five years, we estimate a two-sided matching model that separates two channels behind the gender funding gap, geographic mobility and sorting across accelerator tiers. Women raise about 60% less than men over five years; the gap concentrates among non-relocating women, is largest at active-childrearing ages, and vanishes for relocators, while the mobility cost is near zero for men. Removing mobility frictions raises women's match quality but not their tier; reaching the high-funding top tier also requires removing the sorting disadvantage that women face. The 2012 JOBS Act eased the legal barrier and capacity grew tenfold, yet the U.S. VC dollar gap still tripled (2011-2020): closing it needs mobility, sorting, and capacity together.

econ.GN↗

arXiv2024

THESAURUS: Contrastive Graph Clustering by Swapping Fused Gromov-Wasserstein Couplings

Graph node clustering is a fundamental unsupervised task. Existing methods typically train an encoder through selfsupervised learning and then apply K-means to the encoder output. Some methods use this clustering result directly as the final assignment, while others initialize centroids based on this initial clustering and then finetune both the encoder and these learnable centroids. However, due to their reliance on K-means, these methods inherit its drawbacks when the cluster separability of encoder output is low, facing challenges from the Uniform Effect and Cluster Assimilation. We summarize three reasons for the low cluster separability in existing methods: (1) lack of contextual information prevents discrimination between similar nodes from different clusters; (2) training tasks are not sufficiently aligned with the downstream clustering task; (3) the cluster information in the graph structure is not appropriately exploited. To address these issues, we propose conTrastive grapH clustEring by SwApping fUsed gRomov-wasserstein coUplingS (THESAURUS). Our method introduces semantic prototypes to provide contextual information, and employs a cross-view assignment prediction pretext task that aligns well with the downstream clustering task. Additionally, it utilizes Gromov-Wasserstein Optimal Transport (GW-OT) along with the proposed prototype graph to thoroughly exploit cluster information in the graph structure. To adapt to diverse real-world data, THESAURUS updates the prototype graph and the prototype marginal distribution in OT by using momentum. Extensive experiments demonstrate that THESAURUS achieves higher cluster separability than the prior art, effectively mitigating the Uniform Effect and Cluster Assimilation issues

cs.LG↗

arXiv2024

MemHunter: Automated and Verifiable Memorization Detection at Dataset-scale in LLMs

Large language models (LLMs) have been shown to memorize and reproduce content from their training data, raising significant privacy concerns, especially with web-scale datasets. Existing methods for detecting memorization are primarily sample-specific, relying on manually crafted or discretely optimized memory-inducing prompts generated on a per-sample basis, which become impractical for dataset-level detection due to the prohibitive computational cost of iterating through all samples. In real-world scenarios, data owners may need to verify whether a susceptible LLM has memorized their dataset, particularly if the LLM may have collected the data from the web without authorization. To address this, we introduce MemHunter, which trains a memory-inducing LLM and employs hypothesis testing to efficiently detect memorization at the dataset level, without requiring sample-specific memory inducing. Experiments on models like Pythia and Llama demonstrate that MemHunter can extract up to 40% more training data than existing methods under constrained time resources and reduce search time by up to 80% when integrated as a plug-in. Crucially, MemHunter is the first method capable of dataset-level memorization detection, providing a critical tool for assessing privacy risks in LLMs powered by large-scale datasets.

cs.CR↗

arXiv2024

Toward Ethical Spatial Analysis: Addressing Endogenous Bias Through Visual Analytics

Spatial analysis can generate both exogenous and endogenous biases, which will lead to ethics issues. Exogenous biases arise from external factors or environments and are unrelated to internal operating mechanisms, while endogenous biases stem from internal processes or technologies. Although much attention has been given to exogenous biases, endogenous biases in spatial analysis have been largely overlooked, and a comprehensive methodology for addressing them is yet to be developed. To tackle this challenge, we propose that visual analytics can play a key role in understanding geographic data and improving the interpretation of analytical results. In this study, we conducted a preliminary investigation using various visualization techniques to explore endogenous biases. Our findings demonstrate the potentials of visual analytics to uncover hidden biases and identify associated issues. Additionally, we synthesized these visualization strategies into a framework that approximates a method for detecting endogenous biases. Through this work, we advocate for the integration of visualization at three critical stages of spatial analysis in order to minimize errors, address ethical concerns, and reduce misinterpretations associated with endogenous biases.

cs.HC↗

arXiv2024

PRECISE: Pre-training Sequential Recommenders with Collaborative and Semantic Information

Real-world recommendation systems commonly offer diverse content scenarios for users to interact with. Considering the enormous number of users in industrial platforms, it is infeasible to utilize a single unified recommendation model to meet the requirements of all scenarios. Usually, separate recommendation pipelines are established for each distinct scenario. This practice leads to challenges in comprehensively grasping users' interests. Recent research endeavors have been made to tackle this problem by pre-training models to encapsulate the overall interests of users. Traditional pre-trained recommendation models mainly capture user interests by leveraging collaborative signals. Nevertheless, a prevalent drawback of these systems is their incapacity to handle long-tail items and cold-start scenarios. With the recent advent of large language models, there has been a significant increase in research efforts focused on exploiting LLMs to extract semantic information for users and items. However, text-based recommendations highly rely on elaborate feature engineering and frequently fail to capture collaborative similarities. To overcome these limitations, we propose a novel pre-training framework for sequential recommendation, termed PRECISE. This framework combines collaborative signals with semantic information. Moreover, PRECISE employs a learning framework that initially models users' comprehensive interests across all recommendation scenarios and subsequently concentrates on the specific interests of target-scene behaviors. We demonstrate that PRECISE precisely captures the entire range of user interests and effectively transfers them to the target interests. Empirical findings reveal that the PRECISE framework attains outstanding performance on both public and industrial datasets.

cs.IR↗