Hyperbolic Contrastive Learning with Entailment for Spatial Transcriptomics
为解决空间转录组学中基因表达预测不准确的问题，提出了一种基于双曲几何和基因-图像蕴含损失的对比学习模型HyCLoST，提高了预测精度。
0 citationsRead paper
Selected work
Representative Papers
Efficient Passive Acoustic Monitoring of Killer Whales Using a Two-Stage Detection and Ecotype Classification Cascade
本文提出了一种两阶段级联模型，用于检测和分类虎鲸声学信号，以解决实时监测和分类问题，特别是在数据不平衡情况下。
0 citationsRead paper
Physics-informed learning for the inverse problem in resonant ultrasound spectroscopy
研究通过物理信息学习方法解决共振超声谱中弹性常数的逆问题，利用低维变量和回归模型进行弹性常数重构。
0 citationsRead paper
B-MIM: Biased Masked Image Modeling for Generalizable Segmentation of Fine-Grained Anatomical Structures
为提高细粒度解剖结构分割的泛化能力，本文提出B-MIM方法，通过减少全局语义对齐来优先局部补丁重建，从而增强3D Swin Transformer编码器捕捉高频形态细节的能力。
0 citationsRead paper
Optimal Small Set Expanders and Their Codes
This work investigates the optimal expansion properties of left-regular bipartite graphs over small vertex subsets. By leveraging combinatorial characteristics such as girth, it provides the first complete characterization of s-optimal small-set expanders, proving their existence for all values of s and establishing a transitive lower bound on neighborhood size for larger sets. Building on this theoretical foundation, the paper constructs efficient error-correcting codes tailored for post-quantum key exchange, simultaneously enhancing both small-set expansion performance and coding-based security. The approach integrates techniques from extremal graph theory, expander graph theory, and code construction, offering both theoretical novelty and practical relevance to cryptographic applications.
0 citationsRead paper
Recent publications
Latest Papers
Hyperbolic Contrastive Learning with Entailment for Spatial Transcriptomics
为解决空间转录组学中基因表达预测不准确的问题，提出了一种基于双曲几何和基因-图像蕴含损失的对比学习模型HyCLoST，提高了预测精度。
0 citationsRead paper
Efficient Passive Acoustic Monitoring of Killer Whales Using a Two-Stage Detection and Ecotype Classification Cascade
本文提出了一种两阶段级联模型，用于检测和分类虎鲸声学信号，以解决实时监测和分类问题，特别是在数据不平衡情况下。
0 citationsRead paper
Physics-informed learning for the inverse problem in resonant ultrasound spectroscopy
研究通过物理信息学习方法解决共振超声谱中弹性常数的逆问题，利用低维变量和回归模型进行弹性常数重构。
0 citationsRead paper
B-MIM: Biased Masked Image Modeling for Generalizable Segmentation of Fine-Grained Anatomical Structures
为提高细粒度解剖结构分割的泛化能力，本文提出B-MIM方法，通过减少全局语义对齐来优先局部补丁重建，从而增强3D Swin Transformer编码器捕捉高频形态细节的能力。
0 citationsRead paper
Optimal Small Set Expanders and Their Codes
This work investigates the optimal expansion properties of left-regular bipartite graphs over small vertex subsets. By leveraging combinatorial characteristics such as girth, it provides the first complete characterization of s-optimal small-set expanders, proving their existence for all values of s and establishing a transitive lower bound on neighborhood size for larger sets. Building on this theoretical foundation, the paper constructs efficient error-correcting codes tailored for post-quantum key exchange, simultaneously enhancing both small-set expansion performance and coding-based security. The approach integrates techniques from extremal graph theory, expander graph theory, and code construction, offering both theoretical novelty and practical relevance to cryptographic applications.
0 citationsRead paper