Xindan Gao

Prof. Dr. Xindan Gao

orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 September 2024 in Remote Sensing
Verified
Journal Article
Open access
PDF file
JHJinqi Han
Jinqi Han
YZYing Zhou
Ying Zhou
XGXindan Gao
Xindan Gao
Yinghui ZhaoYinghui Zhao
Yinghui Zhao
Thin clouds in Remote Sensing (RS) imagery can negatively impact subsequent applications. Current Deep Learning (DL) approaches often prioritize information recovery in cloud-covered areas but may not adequately preserve information in cloud-free regions, leading to color distortion, detail loss, and visual artifacts. This study proposes a Sparse Transformer-based Generative Adversarial Network (SpT-GAN) to solve these problems. First, a global enhancement feature extraction module is added to the generator’s top layer to enhance the model’s ability to preserve ground feature information in cloud-free areas. Then, the processed feature map is reconstructed using the sparse transformer-based encoder and decoder with an adaptive threshold filtering mechanism to ensure sparsity. This mechanism enables that the model preserves robust long-range modeling capabilities while disregarding irrelevant details. In addition, inverted residual Fourier transformation blocks are added at each level of the structure to filter redundant information and enhance the quality of the generated cloud-free images. Finally, a composite loss function is created to minimize error in the generated images, resulting in improved resolution and color fidelity. SpT-GAN achieves outstanding results in removing clouds both quantitatively and visually, with Structural Similarity Index (SSIM) values of 98.06% and 92.19% and Peak Signal-to-Noise Ratio (PSNR) values of 36.19 dB and 30.53 dB on the RICE1 and T-Cloud datasets, respectively. On the T-Cloud dataset, especially with more complex cloud components, the superior ability of SpT-GAN to restore ground details is more evident.

14 October 2025 in Journal of King Saud University Computer and Information Sciences
Verified
Journal Article
Open access
PDF file
YZYiqiang Zhou
Yiqiang Zhou
XGXindan Gao
Xindan Gao
JGJifeng Guo
Jifeng Guo
GLGuang Li
Guang Li
LWLu Wang
Lu Wang
JLJing Liu
Jing Liu
Low-light image enhancement aims to improve the quality of images captured under low-light conditions. In previous research, traditional methods and deep learning methods based on Retinex theory have achieved commendable outcomes. However, they often neglect the degradation in low-light images and lack extraction of illumination features; standard attention mechanisms and feed-forward networks also suffer from bottlenecks in computational efficiency and redundant feature removal. To address these issues, we propose LUminance-guided RetineX-based Transformer (LUXFormer), which efficiently removes image degradation while achieving natural and realistic enhancement results. Firstly, we design a spatial-frequency feature enhancer to extract and fuse illumination features, achieving high-quality illumination modeling. Secondly, we propose an illumination-guided sparse attention to address the computational complexity bottleneck of standard self-attention mechanisms through a Top-k selection strategy. Finally, we introduce a refined feed-forward network to further enhance the model’s ability to remove redundant information and capture key features. Extensive experiments on multiple datasets demonstrate that our model outperforms existing mainstream methods. Specifically, our method surpasses the latest state-of-the-art method on the LOLv1 dataset by 2.20dB in PSNR, 0.008 in SSIM, and 0.007 in LPIPS. Furthermore, results in the low-light object detection show that our method significantly improves the performance of downstream tasks, outperforming existing methods on the ExDark dataset with a mean Average Precision (mAP) of 77.26%, demonstrating its superiority and competitiveness.

14 December 2020 in Frontiers in Physiology
Verified
Journal Article
Open access
PDF file
XGXindan Gao
Xindan Gao
CHCraig S. Henriquez
Craig S. Henriquez
WYWenjun Ying
Wenjun Ying
The bidomain equations have been widely used to model the electrical activity of cardiac tissue. While it is well-known that implicit methods have much better stability than explicit methods, implicit methods usually require the solution of a very large nonlinear system of equations at each timestep which is computationally prohibitive. In this work, we present two fully implicit time integration methods for the bidomain equations: the backward Euler method and a second-order one-step two-stage composite backward differentiation formula (CBDF2) which is an L-stable time integration method. Using the backward Euler method as fundamental building blocks, the CBDF2 scheme is easily implementable. After solving the nonlinear system resulting from application of the above two fully implicit schemes by a nonlinear elimination method, the obtained nonlinear global system has a much smaller size, whose Jacobian is symmetric and possibly positive definite. Thus, the residual equation of the approximate Newton approach for the global system can be efficiently solved by standard optimal solvers. As an alternative, we point out that the above two implicit methods combined with operator splittings can also efficiently solve the bidomain equations. Numerical results show that the CBDF2 scheme is an efficient time integration method while achieving high stability and accuracy.

28 March 2025 in Electronics
Verified
Journal Article
Open access
PDF file
XGXindan Gao
Xindan Gao
XBXinyi Ba
Xinyi Ba
JXJian Xing
Jian Xing
YLYing Liu
Ying Liu
Ensuring privacy protection in machine learning is crucial for handling sensitive information, particularly in ethical case studies within computer engineering. Traditional information extraction methods often expose private data to risks such as membership inference and reconstruction attacks, compromising confidentiality. To address these concerns, we propose ChatGLM-LtMP, a privacy-preserving information extraction framework that integrates Least-to-Most Prompting and P-Tuning v2 for structured and secure data retrieval. By employing controlled prompting mechanisms, our approach minimizes data exposure while maintaining high accuracy (93.71%), outperforming baseline models. Additionally, we construct a knowledge graph using the Neo4j 4.4 database and integrate LangChain 0.2 for case-based intelligent question answering. This framework enables secure and interpretable extraction of ethical case data, making it suitable for applications in sensitive machine learning scenarios. The proposed method advances information extraction, safeguarding sensitive ethical cases from potential attacks in automated learning environments.

1 March 2021 in Journal of Physics: Conference Series
Verified
Conference Paper
Open access
PDF file
ACAissata Damaro Camara
Aissata Damaro Camara
XGXindan Gao
Xindan Gao
SOGUIFRET SARL’s service is a full import and export support service for products all over the world. It offers the customer the customs clearance of containers, packaging of effects, shipment of products and objects. The customer files of Soguifret SARL have been exhausted over time due to the large number of customers requesting it at any time, which has led to poor management, as it is a system paper to record customer and staff information. This situation has led to many problems: in particular, unnecessary duplication of data on internal and external customers; loss of information can also occur because data is stored manually and is difficult to locate; therefore, it is difficult to trace the flow of customer information after years of service. The method we have chosen in this article is the Levenshtein distance. He learns the “proofs” by calculating the minimum distance to transform one chain into another and vice versa. In particular, it facilitates data management by avoiding duplication of words in essential documents, gives staff efficiency by saving time and multitasking at the same time. Based on the research results, we have implemented a system using Java; which will facilitate your management of company data.

1 January 2019 in International Journal for Uncertainty Quantification
Verified
Journal Article
Closed access
PDF file
XGXindan Gao
Xindan Gao
WYWenjun Ying
Wenjun Ying
ZZZhiwen Zhang
Zhiwen Zhang
Mathematical models of cardiology involve conductivity and massive parameters describing the dynamics of ionic channels. The conductivity is space dependent and cannot be measured directly. The dynamics of ionic channels are highly nonlinear and the parameters have unavoidable uncertainties since they are estimated using repeated experimental data. Such uncertainties can impact model dependability and credibility since they spread to model parameters during model calibration. It is necessary to study how the uncertainties influence the solution compared to the deterministic solution and to quantify the difference resulting from uncertainty. In this paper, the generalized polynomial chaos method and stochastic collocation method are used to solve the corresponding stochastic partial differential equations. Numerical results are shown to demonstrate that each parameter has different effects on the model responses. More importantly, a quadratic convergence of the expectation is exhibited in the numerical results. The amplitude of standard variance of the stochastic solution can be controlled by the parameter uncertainty. More precisely, the standard variance of the stochastic solution is positively linear to the standard variance of the random parameter. We utilized mono-domain equations which are representative mathematical models to demonstrate the results with the most widely used ionic models Hodgkin-Huxley model and Fitz-Hugh Nagumo model.