arXiv subjects

Yong He

Publications and source records attributed to Yong He.

At least 37 records · Page 2Linked to original sources

arXiv2025

Embodied Intelligence for 3D Understanding: A Survey on 3D Scene Question Answering

3D Scene Question Answering (3D SQA) represents an interdisciplinary task that integrates 3D visual perception and natural language processing, empowering intelligent agents to comprehend and interact with complex 3D environments. Recent advances in large multimodal modelling have driven the creation of diverse datasets and spurred the development of instruction-tuning and zero-shot methods for 3D SQA. However, this rapid progress introduces challenges, particularly in achieving unified analysis and comparison across datasets and baselines. In this survey, we provide the first comprehensive and systematic review of 3D SQA. We organize existing work from three perspectives: datasets, methodologies, and evaluation metrics. Beyond basic categorization, we identify shared architectural patterns across methods. Our survey further synthesizes core limitations and discusses how current trends, such as instruction tuning, multimodal alignment, and zero-shot, can shape future developments. Finally, we propose a range of promising research directions covering dataset construction, task generalization, interaction modeling, and unified evaluation protocols. This work aims to serve as a foundation for future research and foster progress toward more generalizable and intelligent 3D SQA systems.

cs.CV↗

arXiv2025

7B Fully Open Source Moxin-LLM/VLM -- From Pretraining to GRPO-based Reinforcement Learning Enhancement

Recently, Large Language Models (LLMs) have undergone a significant transformation, marked by a rapid rise in both their popularity and capabilities. Leading this evolution are proprietary LLMs like GPT-4 and GPT-o1, which have captured widespread attention in the AI community due to their remarkable performance and versatility. Simultaneously, open-source LLMs, such as LLaMA, have made great contributions to the ever-increasing popularity of LLMs due to the ease to customize and deploy the models across diverse applications. Although open-source LLMs present unprecedented opportunities for innovation and research, the commercialization of LLMs has raised concerns about transparency, reproducibility, and safety. Many open-source LLMs fail to meet fundamental transparency requirements by withholding essential components like training code and data, which may hinder further innovations on LLMs. To mitigate this issue, we introduce Moxin 7B, a fully open-source LLM developed, adhering to principles of open science, open source, open data, and open access. We release the pre-training code and configurations, training and fine-tuning datasets, and intermediate and final checkpoints, aiming to make continuous commitments to fully open-source LLMs. After pre-training the base model, we finetune the Moxin Base model with SOTA post-training framework and instruction data to obtain Moxin Instruct model. To improve the reasoning capability, we further finetune our Instruct model with chain-of-thought data distilled from DeepSeek R1, and then use Group Relative Policy Optimization (GRPO) following DeepSeek R1 to finetune our model, leading to the Moxin Reasoning model. Moreover, we develop our vision language model based on our Moxin model. Experiments show that our models achieve superior performance in various evaluations such as zero-shot evaluation, few-shot evaluation, and CoT evaluation.

cs.CL↗

arXiv2025

Mean uniformly stable function and its application to almost sure stability analysis of randomly switched time-varying systems

This paper investigates uniform almost sure stability of randomly switched time-varying systems. Mode-dependent indefinite multiple Lyapunov functions (iMLFs) are introduced to assess stability properties of diverse time-varying subsystems. To realize the stability conditions establishment based on iMLFs, we present a novel condition so-called mean uniformly stable function for time-varying parameters of iMLFs' derivatives. Our approach provides a probabilistic perspective, making iMLFs well-suited for randomly switched time-varying systems. Moreover, the MUSF condition reveals an essential insight: ensuring that each time-varying subsystem remains mean-bounded during its corresponding sojourn time interval is a prerequisite for the almost sure stability of the entire system. Additionally, the combination of iMLFs and MUSFs is able to accommodate stability analysis scenarios where some subsystems are unstable or exhibit non-exponential decay. Numerical examples are provided to demonstrate the effectiveness and advantages of our approach.

math.OC↗

arXiv2025

Depth as Points: Center Point-based Depth Estimation

The perception of vehicles and pedestrians in urban scenarios is crucial for autonomous driving. This process typically involves complicated data collection, imposes high computational and hardware demands. To address these limitations, we first develop a highly efficient method for generating virtual datasets, which enables the creation of task- and scenario-specific datasets in a short time. Leveraging this method, we construct the virtual depth estimation dataset VirDepth, a large-scale, multi-task autonomous driving dataset. Subsequently, we propose CenterDepth, a lightweight architecture for monocular depth estimation that ensures high operational efficiency and exhibits superior performance in depth estimation tasks with highly imbalanced height-scale distributions. CenterDepth integrates global semantic information through the innovative Center FC-CRFs algorithm, aggregates multi-scale features based on object key points, and enables detection-based depth estimation of targets. Experiments demonstrate that our proposed method achieves superior performance in terms of both computational speed and prediction accuracy.

cs.CV↗

arXiv2025

Penalized Principal Component Analysis for Large-dimension Factor Model with Group Pursuit

This paper investigates the intrinsic group structures within the framework of large-dimensional approximate factor models, which portrays homogeneous effects of the common factors on the individuals that fall into the same group. To this end, we propose a fusion Penalized Principal Component Analysis (PPCA) method and derive a closed-form solution for the $\ell_2$-norm optimization problem. We also show the asymptotic properties of our proposed PPCA estimates. With the PPCA estimates as an initialization, we identify the unknown group structure by a combination of the agglomerative hierarchical clustering algorithm and an information criterion. Then the factor loadings and factor scores are re-estimated conditional on the identified latent groups. Under some regularity conditions, we establish the consistency of the membership estimators as well as that of the group number estimator derived from the information criterion. Theoretically, we show that the post-clustering estimators for the factor loadings and factor scores with group pursuit achieve efficiency gains compared to the estimators by conventional PCA method. Thorough numerical studies validate the established theory and a real financial example illustrates the practical usefulness of the proposed method.

stat.ME↗

arXiv2025

PointDiffuse: A Dual-Conditional Diffusion Model for Enhanced Point Cloud Semantic Segmentation

Diffusion probabilistic models are traditionally used to generate colors at fixed pixel positions in 2D images. Building on this, we extend diffusion models to point cloud semantic segmentation, where point positions also remain fixed, and the diffusion model generates point labels instead of colors. To accelerate the denoising process in reverse diffusion, we introduce a noisy label embedding mechanism. This approach integrates semantic information into the noisy label, providing an initial semantic reference that improves the reverse diffusion efficiency. Additionally, we propose a point frequency transformer that enhances the adjustment of high-level context in point clouds. To reduce computational complexity, we introduce the position condition into MLP and propose denoising PointNet to process the high-resolution point cloud without sacrificing geometric details. Finally, we integrate the proposed noisy label embedding, point frequency transformer and denoising PointNet in our proposed dual conditional diffusion model-based network (PointDiffuse) to perform large-scale point cloud semantic segmentation. Extensive experiments on five benchmarks demonstrate the superiority of PointDiffuse, achieving the state-of-the-art mIoU of 74.2\% on S3DIS Area 5, 81.2\% on S3DIS 6-fold and 64.8\% on SWAN dataset.

cs.CV↗

arXiv2025

TransPCA for Large-dimensional Factor Analysis with Weak Factors: Power Enhancement via Knowledge Transfer

Early work established convergence of the principal component estimators of the factors and loadings up to a rotation for large dimensional approximate factor models with weak factors in that the factor loading $Λ^{(0)}$ scales sublinearly in the number $N$ of cross-section units, i.e., $Λ^{(0)\top}Λ^{(0)}/N^α$ is positive definite in the limit for some $α\in (0,1)$. However, the established convergence rates for weak factors can be much slower especially for small $α$. This article proposes a Transfer Principal Component Analysis (TransPCA) method for enhancing the convergence rates for weak factors by transferring knowledge from large number of available informative panel datasets, which should not be turned a blind eye on in this big data era. We aggregate useful information by analyzing a weighted average projection matrix of the estimated loading spaces from all informative datasets which is highly flexible and computationally efficient. Theoretically, we derive the convergence rates of the estimators of weak/strong loading spaces and factor scores. The results indicate that as long as the auxiliary datasets are similar enough to the target dataset and the auxiliary sample size is sufficiently large, TransPCA estimators can achieve faster convergence rates in contrast to performing PCA solely on the target dataset. To avoid negative transfer, we also investigate the case that the informative datasets are unknown and provide a criterion for selecting useful datasets. Thorough simulation studies and {empirical analysis on real datasets in areas of macroeconomic and finance} are conducted to illustrate the usefulness of our proposed methods where large number of source panel datasets are naturally available.

math.ST↗

arXiv2025

TReND: Transformer derived features and Regularized NMF for neonatal functional network Delineation

Precise parcellation of functional networks (FNs) of early developing human brain is the fundamental basis for identifying biomarker of developmental disorders and understanding functional development. Resting-state fMRI (rs-fMRI) enables in vivo exploration of functional changes, but adult FN parcellations cannot be directly applied to the neonates due to incomplete network maturation. No standardized neonatal functional atlas is currently available. To solve this fundamental issue, we propose TReND, a novel and fully automated self-supervised transformer-autoencoder framework that integrates regularized nonnegative matrix factorization (RNMF) to unveil the FNs in neonates. TReND effectively disentangles spatiotemporal features in voxel-wise rs-fMRI data. The framework integrates confidence-adaptive masks into transformer self-attention layers to mitigate noise influence. A self supervised decoder acts as a regulator to refine the encoder's latent embeddings, which serve as reliable temporal features. For spatial coherence, we incorporate brain surface-based geodesic distances as spatial encodings along with functional connectivity from temporal features. The TReND clustering approach processes these features under sparsity and smoothness constraints, producing robust and biologically plausible parcellations. We extensively validated our TReND framework on three different rs-fMRI datasets: simulated, dHCP and HCP-YA against comparable traditional feature extraction and clustering techniques. Our results demonstrated the superiority of the TReND framework in the delineation of neonate FNs with significantly better spatial contiguity and functional homogeneity. Collectively, we established TReND, a novel and robust framework, for neonatal FN delineation. TReND-derived neonatal FNs could serve as a neonatal functional atlas for perinatal populations in health and disease.

q-bio.NC↗

arXiv2025

Statistical inference for large-dimensional tensor factor model by iterative projections

Tensor Factor Models (TFM) are appealing dimension reduction tools for high-order large-dimensional tensor time series, and have wide applications in economics, finance and medical imaging. In this paper, we propose a projection estimator for the Tucker-decomposition based TFM, and provide its least-square interpretation which parallels to the least-square interpretation of the Principal Component Analysis (PCA) for the vector factor model. The projection technique simultaneously reduces the dimensionality of the signal component and the magnitudes of the idiosyncratic component tensor, thus leading to an increase of the signal-to-noise ratio. We derive a convergence rate of the projection estimator of the loadings and the common factor tensor which are faster than that of the naive PCA-based estimator. Our results are obtained under mild conditions which allow the idiosyncratic components to be weakly cross- and auto- correlated. We also provide a novel iterative procedure based on the eigenvalue-ratio principle to determine the factor numbers. Extensive numerical studies are conducted to investigate the empirical performance of the proposed projection estimators relative to the state-of-the-art ones.

stat.ME↗

arXiv2025

Factor Modelling for Biclustering Large-dimensional Matrix-valued Time Series

A novel unsupervised learning method is proposed in this paper for biclustering large-dimensional matrix-valued time series based on an entirely new latent two-way factor structure. Each block cluster is characterized by its own row and column cluster-specific factors in addition to some common matrix factors which impact on all the matrix time series. We first estimate the global loading spaces by projecting the observation matrices onto the row or column loading space corresponding to common factors. The loading spaces for cluster-specific factors are then further recovered by projecting the observation matrices onto the orthogonal complement space of the estimated global loading spaces. To identify the latent row/column clusters simultaneously for matrix-valued time series, we provide a $K$-means algorithm based on the estimated row/column factor loadings of the cluster-specific weak factors. Theoretically, we derive faster convergence rates for global loading matrices than those of the state-of-the-art methods available in the literature under mild conditions. We also propose an one-pass eigenvalue-ratio method to estimate the numbers of global and cluster-specific factors. The consistency with explicit convergence rates is also established for the estimators of the local loading matrices, the factor numbers and the latent cluster memberships. Numerical experiments with both simulated data as well as a real data example are also reported to illustrate the usefulness of our proposed method.

stat.ME↗

arXiv2025

Integrating remote sensing data assimilation, deep learning and large language model for interactive wheat breeding yield prediction

Yield is one of the core goals of crop breeding. By predicting the potential yield of different breeding materials, breeders can screen these materials at various growth stages to select the best performing. Based on unmanned aerial vehicle remote sensing technology, high-throughput crop phenotyping data in breeding areas is collected to provide data support for the breeding decisions of breeders. However, the accuracy of current yield predictions still requires improvement, and the usability and user-friendliness of yield forecasting tools remain suboptimal. To address these challenges, this study introduces a hybrid method and tool for crop yield prediction, designed to allow breeders to interactively and accurately predict wheat yield by chatting with a large language model (LLM). First, the newly designed data assimilation algorithm is used to assimilate the leaf area index into the WOFOST model. Then, selected outputs from the assimilation process, along with remote sensing inversion results, are used to drive the time-series temporal fusion transformer model for wheat yield prediction. Finally, based on this hybrid method and leveraging an LLM with retrieval augmented generation technology, we developed an interactive yield prediction Web tool that is user-friendly and supports sustainable data updates. This tool integrates multi-source data to assist breeding decision-making. This study aims to accelerate the identification of high-yield materials in the breeding process, enhance breeding efficiency, and enable more scientific and smart breeding decisions.

cs.LG↗

arXiv2024

Representational Transfer Learning for Matrix Completion

We propose to transfer representational knowledge from multiple sources to a target noisy matrix completion task by aggregating singular subspaces information. Under our representational similarity framework, we first integrate linear representation information by solving a two-way principal component analysis problem based on a properly debiased matrix-valued dataset. After acquiring better column and row representation estimators from the sources, the original high-dimensional target matrix completion problem is then transformed into a low-dimensional linear regression, of which the statistical efficiency is guaranteed. A variety of extensional arguments, including post-transfer statistical inference and robustness against negative transfer, are also discussed alongside. Finally, extensive simulation results and a number of real data cases are reported to support our claims.

stat.ML↗

arXiv2024

Generalized Principal Component Analysis for Large-dimensional Matrix Factor Model

Matrix factor models have been growing popular dimension reduction tools for large-dimensional matrix time series. However, the heteroscedasticity of the idiosyncratic components has barely received any attention. Starting from the pseudo likelihood function, this paper introduces a Generalized Principal Component Analysis (GPCA) method for matrix factor model which takes the heteroscedasticity into account. Theoretically, we first derive the asymptotic distributions of the GPCA estimators by assuming the separable covariance matrices are known in advance. We then propose adaptive thresholding estimators for the separable covariance matrices and derive their convergence rates, which is of independent interest. We also show that this would not alter the asymptotic distributions of the GPCA estimators under certain regular sparsity conditions in the high-dimensional covariance matrix estimation literature. The GPCA estimators are shown to be more efficient than the state-of-the-art methods under certain heteroscedasticity conditions. Thorough numerical studies are conducted to demonstrate the superiority of our method over the existing approaches. Analysis of a financial portfolio dataset illustrates the empirical usefulness of the proposed method.

math.ST↗

arXiv2024

Multimodal large language model for wheat breeding: a new exploration of smart breeding

UAV remote sensing technology has become a key technology in crop breeding, which can achieve high-throughput and non-destructive collection of crop phenotyping data. However, the multidisciplinary nature of breeding has brought technical barriers and efficiency challenges to knowledge mining. Therefore, it is important to develop a smart breeding goal tool to mine cross-domain multimodal data. Based on different pre-trained open-source multimodal large language models (MLLMs) (e.g., Qwen-VL, InternVL, Deepseek-VL), this study used supervised fine-tuning (SFT), retrieval-augmented generation (RAG), and reinforcement learning from human feedback (RLHF) technologies to inject cross-domain knowledge into MLLMs, thereby constructing multiple multimodal large language models for wheat breeding (WBLMs). The above WBLMs were evaluated using the newly created evaluation benchmark in this study. The results showed that the WBLM constructed using SFT, RAG and RLHF technologies and InternVL2-8B has leading performance. Then, subsequent experiments were conducted using the WBLM. Ablation experiments indicated that the combination of SFT, RAG, and RLHF technologies can improve the overall generation performance, enhance the generated quality, balance the timeliness and adaptability of the generated answer, and reduce hallucinations and biases. The WBLM performed best in wheat yield prediction using cross-domain data (remote sensing, phenotyping, weather, germplasm) simultaneously, with R2 and RMSE of 0.821 and 489.254 kg/ha, respectively. Furthermore, the WBLM can generate professional decision support answers for phenotyping estimation, environmental stress assessment, target germplasm screening, cultivation technique recommendation, and seed price query tasks.

cs.LG↗

arXiv2024

Biomass phenotyping of oilseed rape through UAV multi-view oblique imaging with 3DGS and SAM model

Biomass estimation of oilseed rape is crucial for optimizing crop productivity and breeding strategies. While UAV-based imaging has advanced high-throughput phenotyping, current methods often rely on orthophoto images, which struggle with overlapping leaves and incomplete structural information in complex field environments. This study integrates 3D Gaussian Splatting (3DGS) with the Segment Anything Model (SAM) for precise 3D reconstruction and biomass estimation of oilseed rape. UAV multi-view oblique images from 36 angles were used to perform 3D reconstruction, with the SAM module enhancing point cloud segmentation. The segmented point clouds were then converted into point cloud volumes, which were fitted to ground-measured biomass using linear regression. The results showed that 3DGS (7k and 30k iterations) provided high accuracy, with peak signal-to-noise ratios (PSNR) of 27.43 and 29.53 and training times of 7 and 49 minutes, respectively. This performance exceeded that of structure from motion (SfM) and mipmap Neural Radiance Fields (Mip-NeRF), demonstrating superior efficiency. The SAM module achieved high segmentation accuracy, with a mean intersection over union (mIoU) of 0.961 and an F1-score of 0.980. Additionally, a comparison of biomass extraction models found the point cloud volume model to be the most accurate, with an determination coefficient (R2) of 0.976, root mean square error (RMSE) of 2.92 g/plant, and mean absolute percentage error (MAPE) of 6.81%, outperforming both the plot crop volume and individual crop volume models. This study highlights the potential of combining 3DGS with multi-view UAV imaging for improved biomass phenotyping.

cs.CV↗

arXiv2024

Simultaneous Estimation and Dataset Selection for Transfer Learning in High Dimensions by a Non-convex Penalty

In this paper, we propose to estimate model parameters and identify informative source datasets simultaneously for high-dimensional transfer learning problems with the aid of a non-convex penalty, in contrast to the separate useful dataset selection and transfer learning procedures in the existing literature. To numerically solve the non-convex problem with respect to two specific statistical models, namely the sparse linear regression and the generalized low-rank trace regression models, we adopt the difference of convex (DC) programming with the alternating direction method of multipliers (ADMM) procedures. We theoretically justify the proposed algorithm from both statistical and computational perspectives. Extensive numerical results are reported alongside to validate the theoretical assertions. An \texttt{R} package \texttt{MHDTL} is developed to implement the proposed methods.

stat.ME↗

arXiv2024

The Devil is in the Sources! Knowledge Enhanced Cross-Domain Recommendation in an Information Bottleneck Perspective

Cross-domain Recommendation (CDR) aims to alleviate the data sparsity and the cold-start problems in traditional recommender systems by leveraging knowledge from an informative source domain. However, previously proposed CDR models pursue an imprudent assumption that the entire information from the source domain is equally contributed to the target domain, neglecting the evil part that is completely irrelevant to users' intrinsic interest. To address this concern, in this paper, we propose a novel knowledge enhanced cross-domain recommendation framework named CoTrans, which remolds the core procedures of CDR models with: Compression on the knowledge from the source domain and Transfer of the purity to the target domain. Specifically, following the theory of Graph Information Bottleneck, CoTrans first compresses the source behaviors with the perception of information from the target domain. Then to preserve all the important information for the CDR task, the feedback signals from both domains are utilized to promote the effectiveness of the transfer procedure. Additionally, a knowledge-enhanced encoder is employed to narrow gaps caused by the non-overlapped items across separate domains. Comprehensive experiments on three widely used cross-domain datasets demonstrate that CoTrans significantly outperforms both single-domain and state-of-the-art cross-domain recommendation approaches.

cs.IR↗

arXiv2024

Deep Learning Based 3D Segmentation: A Survey

3D segmentation is a fundamental and challenging problem in computer vision with applications in autonomous driving and robotics. It has received significant attention from the computer vision, graphics and machine learning communities. Conventional methods for 3D segmentation, based on hand-crafted features and machine learning classifiers, lack generalization ability. Driven by their success in 2D computer vision, deep learning techniques have recently become the tool of choice for 3D segmentation tasks. This has led to an influx of many methods in the literature that have been evaluated on different benchmark datasets. Whereas survey papers on RGB-D and point cloud segmentation exist, there is a lack of a recent in-depth survey that covers all 3D data modalities and application domains. This paper fills the gap and comprehensively surveys the recent progress in deep learning-based 3D segmentation techniques. We cover over 220 works from the last six years, analyze their strengths and limitations, and discuss their competitive results on benchmark datasets. The survey provides a summary of the most commonly used pipelines and finally highlights promising research directions for the future.

cs.CV↗