CFB-GBM v2.0: An Augmented Longitudinal Dataset for Multi-Modal Glioblastoma Segmentation, Radiomics, and RANO Progression Tracking
该研究通过使用预训练的nnU-Net模型完成GTV标注，提高了CFB-GBM数据集在多模态胶质母细胞瘤分割、放射组学及RANO进展跟踪中的完整性与实用性。
0 citationsRead paper
Selected work
Representative Papers
Generalized random forest for extreme quantile regression
Traditional quantile regression methods fail in extreme quantile regression due to sparse tail data and the highly nonlinear, complex structure of conditional quantile functions. To address this, we propose a novel method integrating extreme value theory with statistical learning: it models the covariate-dependent generalized extreme value (GEV) distribution using generalized random forests (GRF) and estimates the conditional tail distribution via the block maxima approach. Our key contribution is the first incorporation of GRF into the extreme quantile regression framework, enabling adaptive, nonparametric modeling of tail characteristics under high-dimensional and heterogeneous covariates. Extensive simulations and empirical analysis on U.S. Fort Collins meteorological data demonstrate that our method substantially outperforms existing quantile regression and extreme-value learning approaches—particularly for quantiles above the 99th percentile, where prediction accuracy improves markedly.
0 citationsRead paper
Task Allocation of UAVs for Monitoring Missions via Hardware-in-the-Loop Simulation and Experimental Validation
To address low task allocation efficiency and the disconnect between theoretical models and actual energy consumption in industrial multi-UAV monitoring, this paper proposes a hybrid optimization framework integrating genetic algorithms with 2-Opt local search, coupled with a hardware-in-the-loop (HIL) simulation platform supporting multi-UAV coordination. By incorporating realistic flight dynamics and high-fidelity battery models, the framework significantly enhances the physical realizability of path planning. Empirical validation demonstrates strong correlation (>0.96) between the optimized theoretical cost function and measured battery depletion and flight time. In real-world industrial deployments, the approach improves task allocation efficiency by 23.5% while maintaining operational feasibility. This work is the first to systematically reveal the strong coupling between abstract task allocation objectives and hardware-level energy metrics, establishing a verifiable modeling and optimization paradigm for physics-aware intelligent UAV scheduling.
0 citationsRead paper
Corecursive Coding of High Computational Derivatives and Power Series
This paper addresses key limitations in higher-order automatic differentiation—namely, susceptibility to truncation error, space explosion, and lack of algebraic closure. We propose a unified computational paradigm based on corecursion and lazy evaluation. Methodologically, we model derivative sequences and formal power series as corecursive data structures and perform symbolic automatic differentiation via abstract syntax tree transformation, enabling on-demand, infinite-length, algebraically closed computation of derivative chains, composite function differentiation, and functional inversion. Our key contribution is the first unified corecursive treatment of both pure derivative streams and power series algebra, thereby eliminating conventional truncation constraints. Experimental results demonstrate substantial reductions in space complexity and combinatorial explosion risk for higher-order differentiation, while correctness and efficiency are validated across multiple analytic functions.
0 citationsRead paper
Recent publications
Latest Papers
CFB-GBM v2.0: An Augmented Longitudinal Dataset for Multi-Modal Glioblastoma Segmentation, Radiomics, and RANO Progression Tracking
该研究通过使用预训练的nnU-Net模型完成GTV标注，提高了CFB-GBM数据集在多模态胶质母细胞瘤分割、放射组学及RANO进展跟踪中的完整性与实用性。
0 citationsRead paper
Generalized random forest for extreme quantile regression
Traditional quantile regression methods fail in extreme quantile regression due to sparse tail data and the highly nonlinear, complex structure of conditional quantile functions. To address this, we propose a novel method integrating extreme value theory with statistical learning: it models the covariate-dependent generalized extreme value (GEV) distribution using generalized random forests (GRF) and estimates the conditional tail distribution via the block maxima approach. Our key contribution is the first incorporation of GRF into the extreme quantile regression framework, enabling adaptive, nonparametric modeling of tail characteristics under high-dimensional and heterogeneous covariates. Extensive simulations and empirical analysis on U.S. Fort Collins meteorological data demonstrate that our method substantially outperforms existing quantile regression and extreme-value learning approaches—particularly for quantiles above the 99th percentile, where prediction accuracy improves markedly.
0 citationsRead paper
Task Allocation of UAVs for Monitoring Missions via Hardware-in-the-Loop Simulation and Experimental Validation
To address low task allocation efficiency and the disconnect between theoretical models and actual energy consumption in industrial multi-UAV monitoring, this paper proposes a hybrid optimization framework integrating genetic algorithms with 2-Opt local search, coupled with a hardware-in-the-loop (HIL) simulation platform supporting multi-UAV coordination. By incorporating realistic flight dynamics and high-fidelity battery models, the framework significantly enhances the physical realizability of path planning. Empirical validation demonstrates strong correlation (>0.96) between the optimized theoretical cost function and measured battery depletion and flight time. In real-world industrial deployments, the approach improves task allocation efficiency by 23.5% while maintaining operational feasibility. This work is the first to systematically reveal the strong coupling between abstract task allocation objectives and hardware-level energy metrics, establishing a verifiable modeling and optimization paradigm for physics-aware intelligent UAV scheduling.
0 citationsRead paper
Corecursive Coding of High Computational Derivatives and Power Series
This paper addresses key limitations in higher-order automatic differentiation—namely, susceptibility to truncation error, space explosion, and lack of algebraic closure. We propose a unified computational paradigm based on corecursion and lazy evaluation. Methodologically, we model derivative sequences and formal power series as corecursive data structures and perform symbolic automatic differentiation via abstract syntax tree transformation, enabling on-demand, infinite-length, algebraically closed computation of derivative chains, composite function differentiation, and functional inversion. Our key contribution is the first unified corecursive treatment of both pure derivative streams and power series algebra, thereby eliminating conventional truncation constraints. Experimental results demonstrate substantial reductions in space complexity and combinatorial explosion risk for higher-order differentiation, while correctness and efficiency are validated across multiple analytic functions.
0 citationsRead paper