$K$-functions for point processes on complex surfaces
本文针对复杂表面上点过程的聚类或规则性评估问题，提出并探讨了几种扩展经典K-函数的方法，其中表现最佳的是基于表面面积的K-函数。
0 citationsRead paper
Selected work
Representative Papers
Hate speech toward migrants on a citizen reporting platform concentrates in neighborhoods undergoing demographic change
This study investigates how immigration influences social inclusion or exclusion in urban neighborhoods at fine-grained spatiotemporal scales, with a focus on the spatial patterns of xenophobic discourse. Leveraging over 550,000 geolocated citizen reports from Chile’s SOSAFE platform and employing a fine-tuned Spanish-language hate speech classifier validated through manual annotation, the research uncovers a “digital boundary” phenomenon: hate speech is not concentrated in traditional immigrant enclaves but instead clusters significantly in neighborhoods where newcomers constituted more than one-third of residents after 2010 and experienced rapid demographic shifts. Conversely, areas with higher educational attainment and fewer recent immigrants emerge as coldspots. The study also reveals that reports referencing immigrants or containing hateful content elicit substantially higher user engagement, underscoring the role of online platforms in reflecting societal tensions.
0 citationsRead paper
Stability and Dual Valuation of Contingent Claims under Rockafellian Perturbations
This study investigates the stability of solutions to discrete-time contingent claims problems under perturbations of both the underlying probability distribution and the claim structure, assuming a finite discrete support. Employing the Rockafellian perturbation framework together with epi-convergence and hypo-convergence from variational analysis, the work establishes—for the first time—a systematic connection between the convergence of primal and dual problems, and uncovers an intrinsic relationship between the duality gap and the value of perfect information. Key contributions include sufficient conditions for strong duality, a proof of solution stability under reasonable perturbations, and the construction of counterexamples that delineate the critical boundary where epi-convergence fails, thereby precisely distinguishing well-posed from ill-posed instances of the problem.
0 citationsRead paper
Agreement coefficients for continuous variables: A review
This study addresses the evaluation of agreement among multiple measurement methods for continuous variables by systematically reviewing and synthesizing mainstream and emerging statistical approaches developed over the past two decades. It encompasses Bland–Altman analysis, Lin’s concordance correlation coefficient, and their extensions to robust, multivariate, repeated-measures, and spatial settings. Notably, the paper introduces probabilistic frameworks and spatial generalizations tailored to modern applications such as image analysis and environmental statistics. By clarifying the historical development, intrinsic connections, and limitations of these methods, the work establishes a unified methodological perspective and delineates promising directions for future research, thereby offering both theoretical grounding and practical guidance for selecting appropriate agreement assessment techniques.
0 citationsRead paper
Optimized imaging prefiltering for enhanced image segmentation
Traditional machine learning models suffer from limited performance in unsupervised or few-shot image segmentation due to insufficient labeled data and suboptimal feature representation. Method: We propose integrating the Box-Cox transformation as a learnable preprocessing module, with a focus on adaptive parameter estimation—replacing fixed or empirically chosen parameters with a statistics-driven algorithm that enhances inter-class separability and feature robustness. Contribution/Results: Experiments demonstrate substantial improvements in discriminant analysis-based segmentation: +8.2% average Dice score and 1.7× inference speedup. In contrast, deep learning models show negligible gains, underscoring the method’s unique efficacy under low-data regimes. To our knowledge, this is the first systematic study revealing how Box-Cox parameter selection differentially impacts segmentation paradigms—highlighting its value for lightweight, interpretable, and data-efficient preprocessing in medical and remote sensing imaging.
0 citationsRead paper
Recent publications
Latest Papers
$K$-functions for point processes on complex surfaces
本文针对复杂表面上点过程的聚类或规则性评估问题，提出并探讨了几种扩展经典K-函数的方法，其中表现最佳的是基于表面面积的K-函数。
0 citationsRead paper
Hate speech toward migrants on a citizen reporting platform concentrates in neighborhoods undergoing demographic change
This study investigates how immigration influences social inclusion or exclusion in urban neighborhoods at fine-grained spatiotemporal scales, with a focus on the spatial patterns of xenophobic discourse. Leveraging over 550,000 geolocated citizen reports from Chile’s SOSAFE platform and employing a fine-tuned Spanish-language hate speech classifier validated through manual annotation, the research uncovers a “digital boundary” phenomenon: hate speech is not concentrated in traditional immigrant enclaves but instead clusters significantly in neighborhoods where newcomers constituted more than one-third of residents after 2010 and experienced rapid demographic shifts. Conversely, areas with higher educational attainment and fewer recent immigrants emerge as coldspots. The study also reveals that reports referencing immigrants or containing hateful content elicit substantially higher user engagement, underscoring the role of online platforms in reflecting societal tensions.
0 citationsRead paper
Stability and Dual Valuation of Contingent Claims under Rockafellian Perturbations
This study investigates the stability of solutions to discrete-time contingent claims problems under perturbations of both the underlying probability distribution and the claim structure, assuming a finite discrete support. Employing the Rockafellian perturbation framework together with epi-convergence and hypo-convergence from variational analysis, the work establishes—for the first time—a systematic connection between the convergence of primal and dual problems, and uncovers an intrinsic relationship between the duality gap and the value of perfect information. Key contributions include sufficient conditions for strong duality, a proof of solution stability under reasonable perturbations, and the construction of counterexamples that delineate the critical boundary where epi-convergence fails, thereby precisely distinguishing well-posed from ill-posed instances of the problem.
0 citationsRead paper
Agreement coefficients for continuous variables: A review
This study addresses the evaluation of agreement among multiple measurement methods for continuous variables by systematically reviewing and synthesizing mainstream and emerging statistical approaches developed over the past two decades. It encompasses Bland–Altman analysis, Lin’s concordance correlation coefficient, and their extensions to robust, multivariate, repeated-measures, and spatial settings. Notably, the paper introduces probabilistic frameworks and spatial generalizations tailored to modern applications such as image analysis and environmental statistics. By clarifying the historical development, intrinsic connections, and limitations of these methods, the work establishes a unified methodological perspective and delineates promising directions for future research, thereby offering both theoretical grounding and practical guidance for selecting appropriate agreement assessment techniques.
0 citationsRead paper
Optimized imaging prefiltering for enhanced image segmentation
Traditional machine learning models suffer from limited performance in unsupervised or few-shot image segmentation due to insufficient labeled data and suboptimal feature representation. Method: We propose integrating the Box-Cox transformation as a learnable preprocessing module, with a focus on adaptive parameter estimation—replacing fixed or empirically chosen parameters with a statistics-driven algorithm that enhances inter-class separability and feature robustness. Contribution/Results: Experiments demonstrate substantial improvements in discriminant analysis-based segmentation: +8.2% average Dice score and 1.7× inference speedup. In contrast, deep learning models show negligible gains, underscoring the method’s unique efficacy under low-data regimes. To our knowledge, this is the first systematic study revealing how Box-Cox parameter selection differentially impacts segmentation paradigms—highlighting its value for lightweight, interpretable, and data-efficient preprocessing in medical and remote sensing imaging.
0 citationsRead paper