arXiv subjects

Jin Liu

Publications and source records attributed to Jin Liu.

At least 55 records · Page 3Linked to original sources

arXiv2026

Vidi2.5: Large Multimodal Models for Video Understanding and Creation

Video has emerged as the primary medium for communication and creativity on the Internet, driving strong demand for scalable, high-quality video production. Vidi models continue to evolve toward next-generation video creation and have achieved state-of-the-art performance in multimodal temporal retrieval (TR). In its second release, Vidi2 advances video understanding with fine-grained spatio-temporal grounding (STG) and extends its capability to video question answering (Video QA), enabling comprehensive multimodal reasoning. Given a text query, Vidi2 can identify not only the corresponding timestamps but also the bounding boxes of target objects within the output time ranges. To enable comprehensive evaluation of STG, we introduce a new benchmark, VUE-STG, which offers critical improvements over existing STG datasets. In addition, we upgrade the previous VUE-TR benchmark to VUE-TR-V2, achieving a more balanced duration and query distribution. Remarkably, the Vidi2 model substantially outperforms leading proprietary systems, such as Gemini 3 Pro Preview and GPT-5, on both VUE-TR-V2 and VUE-STG, while achieving competitive results with popular open-source models with similar scale on video QA benchmarks. The latest Vidi2.5 offers significantly stronger STG capability and slightly better TR and Video QA performance over Vidi2. This update also introduces a Vidi2.5-Think model to handle plot understanding with complex plot reasoning. To comprehensively evaluate the performance of plot understanding, we propose VUE-PLOT benchmark with two tracks, Character and Reasoning. Notably, Vidi2.5-Think outperforms Gemini 3 Pro Preview on fine-grained character understanding with comparable performance on complex plot reasoning. Furthermore, we demonstrate the effectiveness of Vidi2.5 on a challenging real-world application, video editing planning.

cs.CV↗

arXiv2026

MiCo: Multiple Instance Learning with Context-Aware Clustering for Whole Slide Image Analysis

Multiple instance learning (MIL) has shown significant promise in histopathology whole slide image (WSI) analysis for cancer diagnosis and prognosis. However, the inherent spatial heterogeneity of WSIs presents critical challenges, as morphologically similar tissue types are often dispersed across distant anatomical regions. Conventional MIL methods struggle to model these scattered tissue distributions and capture cross-regional spatial interactions effectively. To address these limitations, we propose a novel Multiple instance learning framework with Context-Aware Clustering (MiCo), designed to enhance cross-regional intra-tissue correlations and strengthen inter-tissue semantic associations in WSIs. MiCo begins by clustering instances to distill discriminative morphological patterns, with cluster centroids serving as semantic anchors. To enhance cross-regional intra-tissue correlations, MiCo employs a Cluster Route module, which dynamically links instances of the same tissue type across distant regions via feature similarity. These semantic anchors act as contextual hubs, propagating semantic relationships to refine instance-level representations. To eliminate semantic fragmentation and strengthen inter-tissue semantic associations, MiCo integrates a Cluster Reducer module, which consolidates redundant anchors while enhancing information exchange between distinct semantic groups. Extensive experiments on two challenging tasks across nine large-scale public cancer datasets demonstrate the effectiveness of MiCo, showcasing its superiority over state-of-the-art methods. The code is available at https://github.com/junjianli106/MiCo.

cs.CV↗

arXiv2026

Average shortest-path length in word-adjacency networks: Chinese versus English

Complex networks provide powerful tools for analyzing and understanding the intricate structures present in various systems, including natural language. Here, we analyze topology of growing word-adjacency networks constructed from Chinese and English literary works written in different periods. Unconventionally, instead of considering dictionary words only, we also include punctuation marks as if they were ordinary words. Our approach is based on two arguments: (1) punctuation carries genuine information related to emotional state, allows for logical grouping of content, provides a pause in reading, and facilitates understanding by avoiding ambiguity, and (2) our previous works have shown that punctuation marks behave like words in a Zipfian analysis and, if considered together with regular words, can improve authorship attribution in stylometric studies. We focus on a functional dependence of the average shortest path length $L(N)$ on a network size $N$ for different epochs and individual novels in their original language as well as for translations of selected novels into the other language. We approximate the empirical results with a growing network model and obtain satisfactory agreement between the two. We also observe that $L(N)$ behaves asymptotically similar for both languages if punctuation marks are included but becomes sizably larger for Chinese if punctuation marks are neglected.

cs.CL↗

arXiv2025

MICCAI STSR 2025 Challenge: Semi-Supervised Teeth and Pulp Segmentation and CBCT-IOS Registration

Cone-Beam Computed Tomography (CBCT) and Intraoral Scanning (IOS) are essential for digital dentistry, but annotated data scarcity limits automated solutions for pulp canal segmentation and cross-modal registration. To benchmark semi-supervised learning (SSL) in this domain, we organized the STSR 2025 Challenge at MICCAI 2025, featuring two tasks: (1) semi-supervised segmentation of teeth and pulp canals in CBCT, and (2) semi-supervised rigid registration of CBCT and IOS. We provided 60 labeled and 640 unlabeled IOS samples, plus 30 labeled and 250 unlabeled CBCT scans with varying resolutions and fields of view. The challenge attracted strong community participation, with top teams submitting open-source deep learning-based SSL solutions. For segmentation, leading methods used nnU-Net and Mamba-like State Space Models with pseudo-labeling and consistency regularization, achieving a Dice score of 0.967 and Instance Affinity of 0.738 on the hidden test set. For registration, effective approaches combined PointNetLK with differentiable SVD and geometric augmentation to handle modality gaps; hybrid neural-classical refinement enabled accurate alignment despite limited labels. All data and code are publicly available at https://github.com/ricoleehduu/STS-Challenge-2025 to ensure reproducibility.

cs.CV↗

arXiv2025

Jenss-Bayley Latent Change Score Model with Individual Ratio of Growth Acceleration in the Framework of Individual Measurement Occasions

Longitudinal data analysis has been widely employed to examine between-individual differences in within-individual changes. One challenge of such analyses is that the rate-of-change is only available indirectly when change patterns are nonlinear with respect to time. Latent change score models (LCSMs), which can be employed to investigate the change in rate-of-change at the individual level, have been developed to address this challenge. We extend an existing LCSM with the Jenss-Bayley growth curve \cite[Chapter~18]{Grimm2016growth} and propose a novel expression for change scores that allows for (1) unequally-spaced study waves and (2) individual measurement occasions around each wave. We also extend the existing model to estimate the individual ratio of the growth acceleration (that largely determines the trajectory shape and is viewed as the most important parameter in the Jenss-Bayley model). We present the proposed model by a simulation study and a real-world data analysis. Our simulation study demonstrates that the proposed model can estimate the parameters unbiasedly and precisely and exhibit target confidence interval coverage. The simulation study also shows that the proposed model with the novel expression for the change scores outperforms the existing model. An empirical example using longitudinal reading scores shows that the model can estimate the individual ratio of the growth acceleration and generate individual rate-of-change in practice. We also provide the corresponding code for the proposed model.

stat.ME↗

arXiv2025

Extending Latent Basis Growth Model to Explore Joint Development in the Framework of Individual Measurement Occasions

Longitudinal processes often pose nonlinear change patterns. Latent basis growth models (LBGMs) provide a versatile solution without requiring specific functional forms. Building on the LBGM specification for unequally-spaced waves and individual occasions proposed by Liu and Perera (2023), we extend LBGMs to multivariate longitudinal outcomes. This provides a unified approach to nonlinear, interconnected trajectories. Simulation studies demonstrate that the proposed model can provide unbiased and accurate estimates with target coverage probabilities for the parameters of interest. Real-world analyses of reading and mathematics scores demonstrates its effectiveness in analyzing joint developmental processes that vary in temporal patterns. Computational code is included.

stat.ME↗

arXiv2025

Estimating Rate of Change for Nonlinear Trajectories in the Framework of Individual Measurement Occasions: A New Perspective on Growth Curves

Researchers are often interested in examining between-individual differences in within-individual processes. If the process under investigation is tracked for a long time, its trajectory may show a certain degree of nonlinearity, so that the rate-of-change is not constant. A fundamental goal of modeling such nonlinear processes is to estimate model parameters that reflect meaningful aspects of change, including the parameters related to change and other parameters that shed light on substantive hypotheses. However, if the measurement occasion is unstructured, existing models cannot simultaneously estimate these two types of parameters. This article has three goals. First, we view the change over time as the area under the curve (AUC) of the rate-of-change versus time (r-t) graph. Second, using the instantaneous rate-of-change midway through a time interval to approximate the average rate-of-change during that interval, we propose a new specification to describe longitudinal processes. In addition to obtaining the individual change-related parameters and other parameters related to specific research questions, the new specification allows for unequally-space study waves and individual measurement occasions around each wave. Third, we derive the model-based interval-specific change and change-from-baseline, two common measures to evaluate change over time. We evaluate the proposed specification through a simulation study and a real-world data analysis. We also provide OpenMx and Mplus 8 code for each model with the novel specification.

stat.ME↗

arXiv2025

Decomposing Impact on Longitudinal Outcome of Time-varying Covariate into Baseline Effect and Temporal Effect

Longitudinal processes are often associated with each other over time; therefore, it is important to investigate the associations among developmental processes and understand their joint development. The latent growth curve model (LGCM) with a time-varying covariate (TVC) provides a method to estimate the TVC's effect on a longitudinal outcome while simultaneously modeling the outcome's change. However, it does not allow the TVC to predict variations in the random growth coefficients. We propose decomposing the TVC's effect into initial trait and temporal states using three methods to address this limitation. In each method, the baseline of the TVC is viewed as an initial trait, and the corresponding effects are obtained by regressing random intercepts and slopes on the baseline value. Temporal states are characterized as (1) interval-specific slopes, (2) interval-specific changes, or (3) changes from the baseline at each measurement occasion, depending on the method. We demonstrate our methods through simulations and real-world data analyses, assuming a linear-linear functional form for the longitudinal outcome. The results demonstrate that LGCMs with a decomposed TVC can provide unbiased and precise estimates with target confidence intervals. We also provide OpenMx and Mplus 8 code for these methods with commonly used linear and nonlinear functions.

stat.ME↗

arXiv2025

Further Exploration of the Effects of Time-varying Covariate in Growth Mixture Models with Nonlinear Trajectories

Growth mixture modeling (GMM) is an analytical tool for identifying multiple unobserved sub-populations of longitudinal processes. In particular, it describes change patterns within each latent sub-population and examines between-individual differences in within-individual change for each sub-group. One research interest in utilizing GMMs is to explore how covariates affect such heterogeneity in change patterns. Liu and Perera (2022c) extended mixture-of-experts (MoE) models, which mainly focus on time-invariant covariates, for allowing the covariates to account for within-group and between-group differences simultaneously and examining the heterogeneity in nonlinear trajectories. The present study further extends Liu and Perera (2022c) and examines the effects on trajectory heterogeneity of time-varying covariates (TVCs). Specifically, we propose methods to decompose a TVC into a trait feature (e.g., the baseline value of the TVC) and a set of state features (e.g., interval-specific slopes or changes). The trait features are allowed to account for within-group differences in growth factors of trajectories (i.e., trait effect), and the state features are allowed to impact observed values of a longitudinal process (i.e., state effect). We examine the proposed models using a simulation study and a real-world data analysis. The simulation study demonstrated that the proposed models are capable of separating trajectories into several clusters and generally generating unbiased and accurate estimates with target coverage probabilities. With the proposed models, we showed the heterogeneity in the trait and state features of reading ability across latent classes of students' mathematics performance. Meanwhile, the trait and state effects on mathematics development of reading ability are also heterogeneous across the clusters of students.

stat.ME↗

arXiv2025

Examination of Nonlinear Longitudinal Processes with Latent Variables, Latent Processes, Latent Changes, and Latent Classes in the Structural Equation Modeling Framework: The R package nlpsem

We introduce the R package nlpsem, a comprehensive toolkit for analyzing longitudinal processes within the structural equation modeling (SEM) framework, incorporating individual measurement occasions. This package emphasizes nonlinear longitudinal models, especially intrinsic ones, across four key scenarios: (1) univariate longitudinal processes with latent variables, optionally including covariates such as time-invariant covariates (TICs) and time-varying covariates (TVCs); (2) multivariate longitudinal analyses to explore correlations or unidirectional relationships between longitudinal variables; (3) multiple-group frameworks for comparing manifest classes in scenarios (1) and (2); and (4) mixture models for scenarios (1) and (2), accommodating latent class heterogeneity. Built on the OpenMx R package, nlpsem supports flexible model designs and uses the full information maximum likelihood method for parameter estimation. A notable feature is its algorithm for determining initial values directly from raw data, enhancing computational efficiency and convergence. Furthermore, nlpsem provides tools for goodness-of-fit tests, cluster analyses, visualization, derivation of p-values and three types of confidence intervals, as well as model selection for nested models using likelihood ratio tests and for non-nested models based on criteria such as Akaike Information Criterion and Bayesian Information Criterion. This article serves as a companion document to the nlpsem R package, providing a comprehensive guide to its modeling capabilities, estimation methods, implementation features, and application examples using synthetic intelligence growth data.

stat.ME↗

arXiv2025

MICCAI STS 2024 Challenge: Semi-Supervised Instance-Level Tooth Segmentation in Panoramic X-ray and CBCT Images

Orthopantomogram (OPGs) and Cone-Beam Computed Tomography (CBCT) are vital for dentistry, but creating large datasets for automated tooth segmentation is hindered by the labor-intensive process of manual instance-level annotation. This research aimed to benchmark and advance semi-supervised learning (SSL) as a solution for this data scarcity problem. We organized the 2nd Semi-supervised Teeth Segmentation (STS 2024) Challenge at MICCAI 2024. We provided a large-scale dataset comprising over 90,000 2D images and 3D axial slices, which includes 2,380 OPG images and 330 CBCT scans, all featuring detailed instance-level FDI annotations on part of the data. The challenge attracted 114 (OPG) and 106 (CBCT) registered teams. To ensure algorithmic excellence and full transparency, we rigorously evaluated the valid, open-source submissions from the top 10 (OPG) and top 5 (CBCT) teams, respectively. All successful submissions were deep learning-based SSL methods. The winning semi-supervised models demonstrated impressive performance gains over a fully-supervised nnU-Net baseline trained only on the labeled data. For the 2D OPG track, the top method improved the Instance Affinity (IA) score by over 44 percentage points. For the 3D CBCT track, the winning approach boosted the Instance Dice score by 61 percentage points. This challenge confirms the substantial benefit of SSL for complex, instance-level medical image segmentation tasks where labeled data is scarce. The most effective approaches consistently leveraged hybrid semi-supervised frameworks that combined knowledge from foundational models like SAM with multi-stage, coarse-to-fine refinement pipelines. Both the challenge dataset and the participants' submitted code have been made publicly available on GitHub (https://github.com/ricoleehduu/STS-Challenge-2024), ensuring transparency and reproducibility.

eess.IV↗

arXiv2025

Wavefront-Constrained Passive Obscured Object Detection

Accurately localizing and segmenting obscured objects from faint light patterns beyond the field of view is highly challenging due to multiple scattering and medium-induced perturbations. Most existing methods, based on real-valued modeling or local convolutional operations, are inadequate for capturing the underlying physics of coherent light propagation. Moreover, under low signal-to-noise conditions, these methods often converge to non-physical solutions, severely compromising the stability and reliability of the observation. To address these challenges, we propose a novel physics-driven Wavefront Propagating Compensation Network (WavePCNet) to simulate wavefront propagation and enhance the perception of obscured objects. This WavePCNet integrates the Tri-Phase Wavefront Complex-Propagation Reprojection (TriWCP) to incorporate complex amplitude transfer operators to precisely constrain coherent propagation behavior, along with a momentum memory mechanism to effectively suppress the accumulation of perturbations. Additionally, a High-frequency Cross-layer Compensation Enhancement is introduced to construct frequency-selective pathways with multi-scale receptive fields and dynamically model structural consistency across layers, further boosting the model's robustness and interpretability under complex environmental conditions. Extensive experiments conducted on four physically collected datasets demonstrate that WavePCNet consistently outperforms state-of-the-art methods across both accuracy and robustness.

cs.CV↗

arXiv2025

Opt3DGS: Optimizing 3D Gaussian Splatting with Adaptive Exploration and Curvature-Aware Exploitation

3D Gaussian Splatting (3DGS) has emerged as a leading framework for novel view synthesis, yet its core optimization challenges remain underexplored. We identify two key issues in 3DGS optimization: entrapment in suboptimal local optima and insufficient convergence quality. To address these, we propose Opt3DGS, a robust framework that enhances 3DGS through a two-stage optimization process of adaptive exploration and curvature-guided exploitation. In the exploration phase, an Adaptive Weighted Stochastic Gradient Langevin Dynamics (SGLD) method enhances global search to escape local optima. In the exploitation phase, a Local Quasi-Newton Direction-guided Adam optimizer leverages curvature information for precise and efficient convergence. Extensive experiments on diverse benchmark datasets demonstrate that Opt3DGS achieves state-of-the-art rendering quality by refining the 3DGS optimization process without modifying its underlying representation.

cs.CV↗

arXiv2025

Where Is Self-admitted Code Generated by Large Language Models on GitHub?

The increasing use of Large Language Models (LLMs) in software development has garnered significant attention from researchers evaluating the capabilities and limitations of LLMs for code generation. However, much of the research focuses on controlled datasets such as HumanEval, which do not adequately capture the characteristics of LLM-generated code in real-world development scenarios. To address this gap, our study investigates self-admitted code generated by LLMs on GitHub, specifically focusing on instances where developers in projects with over five stars acknowledge the use of LLMs to generate code through code comments. Our findings reveal several key insights: (1) ChatGPT and Copilot dominate code generation, with minimal contributions from other LLMs. (2) Projects containing ChatGPT/Copilot-generated code appears in small/medium-sized projects led by small teams, which are continuously evolving. (3) ChatGPT/Copilot-generated code generally is a minor project portion, primarily generating short/moderate-length, low-complexity snippets (e.g., algorithms and data structures code; text processing code). (4) ChatGPT/Copilot-generated code generally undergoes minimal modifications, with bug-related changes ranging from 4% to 12%. (5) Most code comments only state LLM use, while few include details like prompts, human edits, or code testing status. Based on these findings, we discuss the implications for researchers and practitioners.

cs.SE↗

arXiv2025

FedMuon: Accelerating Federated Learning with Matrix Orthogonalization

The core bottleneck of Federated Learning (FL) lies in the communication rounds. That is, how to achieve more effective local updates is crucial for reducing communication rounds. Existing FL methods still primarily use element-wise local optimizers (Adam/SGD), neglecting the geometric structure of the weight matrices. This often leads to the amplification of pathological directions in the weights during local updates, leading deterioration in the condition number and slow convergence. Therefore, we introduce the Muon optimizer in local, which has matrix orthogonalization to optimize matrix-structured parameters. Experimental results show that, in IID setting, Local Muon significantly accelerates the convergence of FL and reduces communication rounds compared to Local SGD and Local AdamW. However, in non-IID setting, independent matrix orthogonalization based on the local distributions of each client induces strong client drift. Applying Muon in non-IID FL poses significant challenges: (1) client preconditioner leading to client drift; (2) moment reinitialization. To address these challenges, we propose a novel Federated Muon optimizer (FedMuon), which incorporates two key techniques: (1) momentum aggregation, where clients use the aggregated momentum for local initialization; (2) local-global alignment, where the local gradients are aligned with the global update direction to significantly reduce client drift. Theoretically, we prove that \texttt{FedMuon} achieves a linear speedup convergence rate without the heterogeneity assumption, where $S$ is the number of participating clients per round, $K$ is the number of local iterations, and $R$ is the total number of communication rounds. Empirically, we validate the effectiveness of FedMuon on language and vision models. Compared to several baselines, FedMuon significantly reduces communication rounds and improves test accuracy.

cs.LG↗

arXiv2025

Highly Undersampled MRI Reconstruction via a Single Posterior Sampling of Diffusion Models

Incoherent k-space undersampling and deep learning-based reconstruction methods have shown great success in accelerating MRI. However, the performance of most previous methods will degrade dramatically under high acceleration factors, e.g., 8$\times$ or higher. Recently, denoising diffusion models (DM) have demonstrated promising results in solving this issue; however, one major drawback of the DM methods is the long inference time due to a dramatic number of iterative reverse posterior sampling steps. In this work, a Single Step Diffusion Model-based reconstruction framework, namely SSDM-MRI, is proposed for restoring MRI images from highly undersampled k-space. The proposed method achieves one-step reconstruction by first training a conditional DM and then iteratively distilling this model four times using an iterative selective distillation algorithm, which works synergistically with a shortcut reverse sampling strategy for model inference. Comprehensive experiments were carried out on both publicly available fastMRI brain and knee images, as well as an in-house multi-echo GRE (QSM) subject. Overall, the results showed that SSDM-MRI outperformed other methods in terms of numerical metrics (e.g., PSNR and SSIM), error maps, image fine details, and latent susceptibility information hidden in MRI phase images. In addition, the reconstruction time for a 320$\times$320 brain slice of SSDM-MRI is only 0.45 second, which is only comparable to that of a simple U-net, making it a highly effective solution for MRI reconstruction tasks.

eess.IV↗

arXiv2025

CEPerFed: Communication-Efficient Personalized Federated Learning for Multi-Pulse MRI Classification

Multi-pulse magnetic resonance imaging (MRI) is widely utilized for clinical practice such as Alzheimer's disease diagnosis. To train a robust model for multi-pulse MRI classification, it requires large and diverse data from various medical institutions while protecting privacy by preventing raw data sharing across institutions. Although federated learning (FL) is a feasible solution to address this issue, it poses challenges of model convergence due to the effect of data heterogeneity and substantial communication overhead due to large numbers of parameters transmitted within the model. To address these challenges, we propose CEPerFed, a communication-efficient personalized FL method. It mitigates the effect of data heterogeneity by incorporating client-side historical risk gradients and historical mean gradients to coordinate local and global optimization. The former is used to weight the contributions from other clients, enhancing the reliability of local updates, while the latter enforces consistency between local updates and the global optimization direction to ensure stable convergence across heterogeneous data distributions. To address the high communication overhead, we propose a hierarchical SVD (HSVD) strategy that transmits only the most critical information required for model updates. Experiments on five classification tasks demonstrate the effectiveness of the CEPerFed method. The code will be released upon acceptance at https://github.com/LD0416/CEPerFed.

cs.LG↗

arXiv2025

A Semiparametric Gaussian Mixture Model with Spatial Dependence and Its Application to Whole-Slide Image Clustering Analysis

We develop here a semiparametric Gaussian mixture model (SGMM) for unsupervised learning with valuable spatial information taken into consideration. Specifically, we assume for each instance a random location. Then, conditional on this random location, we assume for the feature vector a standard Gaussian mixture model (GMM). The proposed SGMM allows the mixing probability to be nonparametrically related to the spatial location. Compared with a classical GMM, SGMM is considerably more flexible and allows the instances from the same class to be spatially clustered. To estimate the SGMM, novel EM algorithms are developed and rigorous asymptotic theories are established. Extensive numerical simulations are conducted to demonstrate our finite sample performance. For a real application, we apply our SGMM method to the CAMELYON16 dataset of whole-slide images (WSIs) for breast cancer detection. The SGMM method demonstrates outstanding clustering performance.

stat.ME↗