Point Cloud Quality for Meshfree Methods
研究针对无网格点云质量缺乏系统性研究的问题，通过比较现有及新提出的质量度量，并经广泛数值测试，确定了六个可靠的点云质量指标。
0 citationsRead paper
Selected work
Representative Papers
Data-driven subsampling rates for diffusion parameter estimation of SDEs
This study addresses the challenge of estimating diffusion parameters in stochastic differential equation (SDE) models when data and model are compatible only at specific scales. The authors propose an adaptive subsampling method based on the statistics of monotonic runs. By demonstrating that, for a broad class of additive-noise SDEs, the length of monotonic runs at infinitesimal scales approximately follows a geometric distribution with success probability 1/2, they establish a general criterion for selecting the subsampling rate without relying on multiscale diffusion asymptotics. The optimal sampling scale matching the SDE’s infinitesimal behavior is automatically determined solely from the statistical properties of monotonic increasing or decreasing segments in the observed time series. Validation on surrogate modeling of fiber lay-down trajectories in nonwoven fabric production demonstrates that the method yields highly accurate and model-consistent diffusion parameter estimates, proving effective in real-world industrial applications.
0 citationsRead paper
Deep Learning for Model Calibration in Simulation of Itaconic Acid Production
This study addresses the challenges of parameter estimation and limited generalizability in kinetic models for itaconic acid fermentation across varying agitation speeds and bioreactor scales. To overcome these limitations, the authors introduce, for the first time, generative conditional flow matching (CFM) into bioprocess modeling, leveraging multi-condition batch experimental data to calibrate model parameters. Compared to conventional nonlinear regression and direct deep learning (DDL) approaches, CFM substantially enhances both parameter estimation accuracy and cross-scale predictive performance. Scale-up experiments demonstrate excellent agreement between CFM-predicted concentration profiles and experimental measurements, confirming the method’s reliability and data efficiency in modeling dynamic biological processes.
0 citationsRead paper
Remote Sensing Imagery for Flood Detection: Exploration of Augmentation Strategies
To address the challenges of limited annotated samples and poor generalization in river flood detection from remote sensing RGB imagery, this work systematically evaluates the impact of three categories of data augmentation—geometric transformations, color perturbations, and optical distortions—on semantic segmentation models (e.g., U-Net). It is the first study to empirically validate the effectiveness of optical distortion augmentation specifically for flood detection and proposes a water-body-aware augmentation selection criterion. Experiments on the BlessemFlood21 dataset demonstrate that the proposed augmentation strategy improves IoU for small-scale flood regions by 12.3%, significantly reduces false positives, and enhances model robustness. The findings yield a reusable, optimization-oriented data augmentation framework tailored for real-time, high-accuracy remote sensing–based flood monitoring.
0 citationsRead paper
Preconditioning Natural and Second Order Gradient Descent in Quantum Optimization: A Performance Benchmark
Addressing the three key challenges in parameterized quantum circuit optimization—non-convex objective landscapes, high gradient noise, and barren plateaus—this work systematically evaluates natural gradient and second-order optimizers for QAOA-based MaxCut solving. We propose SP-BFGS: a robust quasi-Newton method that integrates secant penalty regularization into the BFGS framework to enhance resilience against gradient noise while preserving convergence stability and computational efficiency. SP-BFGS combines quantum natural gradient estimation, BFGS-type Hessian approximation, secant-constrained regularization, and shallow-depth QAOA simulation. On synthetic MaxCut instances, SP-BFGS achieves significantly faster convergence and higher-quality solutions compared to standard BFGS and Adam, demonstrating its effectiveness and practicality for noisy intermediate-scale quantum (NISQ) optimization.
0 citationsRead paper
Recent publications
Latest Papers
Point Cloud Quality for Meshfree Methods
研究针对无网格点云质量缺乏系统性研究的问题，通过比较现有及新提出的质量度量，并经广泛数值测试，确定了六个可靠的点云质量指标。
0 citationsRead paper
Data-driven subsampling rates for diffusion parameter estimation of SDEs
This study addresses the challenge of estimating diffusion parameters in stochastic differential equation (SDE) models when data and model are compatible only at specific scales. The authors propose an adaptive subsampling method based on the statistics of monotonic runs. By demonstrating that, for a broad class of additive-noise SDEs, the length of monotonic runs at infinitesimal scales approximately follows a geometric distribution with success probability 1/2, they establish a general criterion for selecting the subsampling rate without relying on multiscale diffusion asymptotics. The optimal sampling scale matching the SDE’s infinitesimal behavior is automatically determined solely from the statistical properties of monotonic increasing or decreasing segments in the observed time series. Validation on surrogate modeling of fiber lay-down trajectories in nonwoven fabric production demonstrates that the method yields highly accurate and model-consistent diffusion parameter estimates, proving effective in real-world industrial applications.
0 citationsRead paper
Deep Learning for Model Calibration in Simulation of Itaconic Acid Production
This study addresses the challenges of parameter estimation and limited generalizability in kinetic models for itaconic acid fermentation across varying agitation speeds and bioreactor scales. To overcome these limitations, the authors introduce, for the first time, generative conditional flow matching (CFM) into bioprocess modeling, leveraging multi-condition batch experimental data to calibrate model parameters. Compared to conventional nonlinear regression and direct deep learning (DDL) approaches, CFM substantially enhances both parameter estimation accuracy and cross-scale predictive performance. Scale-up experiments demonstrate excellent agreement between CFM-predicted concentration profiles and experimental measurements, confirming the method’s reliability and data efficiency in modeling dynamic biological processes.
0 citationsRead paper
Remote Sensing Imagery for Flood Detection: Exploration of Augmentation Strategies
To address the challenges of limited annotated samples and poor generalization in river flood detection from remote sensing RGB imagery, this work systematically evaluates the impact of three categories of data augmentation—geometric transformations, color perturbations, and optical distortions—on semantic segmentation models (e.g., U-Net). It is the first study to empirically validate the effectiveness of optical distortion augmentation specifically for flood detection and proposes a water-body-aware augmentation selection criterion. Experiments on the BlessemFlood21 dataset demonstrate that the proposed augmentation strategy improves IoU for small-scale flood regions by 12.3%, significantly reduces false positives, and enhances model robustness. The findings yield a reusable, optimization-oriented data augmentation framework tailored for real-time, high-accuracy remote sensing–based flood monitoring.
0 citationsRead paper
Preconditioning Natural and Second Order Gradient Descent in Quantum Optimization: A Performance Benchmark
Addressing the three key challenges in parameterized quantum circuit optimization—non-convex objective landscapes, high gradient noise, and barren plateaus—this work systematically evaluates natural gradient and second-order optimizers for QAOA-based MaxCut solving. We propose SP-BFGS: a robust quasi-Newton method that integrates secant penalty regularization into the BFGS framework to enhance resilience against gradient noise while preserving convergence stability and computational efficiency. SP-BFGS combines quantum natural gradient estimation, BFGS-type Hessian approximation, secant-constrained regularization, and shallow-depth QAOA simulation. On synthetic MaxCut instances, SP-BFGS achieves significantly faster convergence and higher-quality solutions compared to standard BFGS and Adam, demonstrating its effectiveness and practicality for noisy intermediate-scale quantum (NISQ) optimization.
0 citationsRead paper