Generalized Graph Search Trees
本文研究了广义图搜索树的识别问题，通过允许每个前驱邻居作为父节点来泛化搜索树概念，并探讨了该问题在不同条件下的复杂性。
0 citationsRead paper
Selected work
Representative Papers
Real-time and adaptive anomaly detection algorithm for cyclostationary models
本文提出PeriodicCALM，一种针对周期平稳数据流的有效实时异常检测框架，通过考虑周期性变化来准确识别真实异常并减少误报。
0 citationsRead paper
Language-Representability: Possibilities and Limitations
研究通过扩展二进制语言框架，探讨了小语言如何描述已知图类，并指出该方法在表征稀疏图（如平面图）时的局限性。
0 citationsRead paper
Real-time anomaly detection in base station testbeds via scalable kernel density estimation framework
This work addresses the challenge of abrupt temporal metric anomalies in large-scale base station testing, often caused by resource allocation errors, which necessitate efficient unsupervised detection methods. The authors propose CALM, a framework that leverages nonparametric kernel density estimation combined with bootstrap-based dynamic thresholding to enable real-time anomaly detection at the individual testbed level. To mitigate alert fatigue and identify globally significant anomalies, they further introduce AggCALM, which aggregates signals across multiple testbeds. Requiring no labeled data, the approach offers high timeliness and scalability, demonstrating strong performance on both simulated and real-world base station datasets. Beyond ensuring stable operation in complex testing environments, the method is readily transferable to other system health monitoring scenarios.
0 citationsRead paper
Simultaneous Graph Parameters and How to Bound Them
This study investigates boundedness relationships between the simultaneous 𝒞-number and classical graph parameters, aiming to characterize which parameters can upper-bound the simultaneous 𝒞-number. By leveraging structural graph theory, parameterized complexity, set-representation techniques, closure properties of graph classes, and function-mapping methods, the work establishes for the first time a systematic equivalence between boundedness of parameters such as cliquewidth and mim-width within the simultaneous 𝒞-number framework—while showing that modular-width does not share this property. It also fully characterizes the graph-class conditions under which parameters like treewidth upper-bound the simultaneous 𝒞-number. As an application, the results yield efficient algorithms for the Clique problem on the corresponding graph classes.
0 citationsRead paper
Recent publications
Latest Papers
Generalized Graph Search Trees
本文研究了广义图搜索树的识别问题，通过允许每个前驱邻居作为父节点来泛化搜索树概念，并探讨了该问题在不同条件下的复杂性。
0 citationsRead paper
Real-time and adaptive anomaly detection algorithm for cyclostationary models
本文提出PeriodicCALM，一种针对周期平稳数据流的有效实时异常检测框架，通过考虑周期性变化来准确识别真实异常并减少误报。
0 citationsRead paper
Language-Representability: Possibilities and Limitations
研究通过扩展二进制语言框架，探讨了小语言如何描述已知图类，并指出该方法在表征稀疏图（如平面图）时的局限性。
0 citationsRead paper
Real-time anomaly detection in base station testbeds via scalable kernel density estimation framework
This work addresses the challenge of abrupt temporal metric anomalies in large-scale base station testing, often caused by resource allocation errors, which necessitate efficient unsupervised detection methods. The authors propose CALM, a framework that leverages nonparametric kernel density estimation combined with bootstrap-based dynamic thresholding to enable real-time anomaly detection at the individual testbed level. To mitigate alert fatigue and identify globally significant anomalies, they further introduce AggCALM, which aggregates signals across multiple testbeds. Requiring no labeled data, the approach offers high timeliness and scalability, demonstrating strong performance on both simulated and real-world base station datasets. Beyond ensuring stable operation in complex testing environments, the method is readily transferable to other system health monitoring scenarios.
0 citationsRead paper
Simultaneous Graph Parameters and How to Bound Them
This study investigates boundedness relationships between the simultaneous 𝒞-number and classical graph parameters, aiming to characterize which parameters can upper-bound the simultaneous 𝒞-number. By leveraging structural graph theory, parameterized complexity, set-representation techniques, closure properties of graph classes, and function-mapping methods, the work establishes for the first time a systematic equivalence between boundedness of parameters such as cliquewidth and mim-width within the simultaneous 𝒞-number framework—while showing that modular-width does not share this property. It also fully characterizes the graph-class conditions under which parameters like treewidth upper-bound the simultaneous 𝒞-number. As an application, the results yield efficient algorithms for the Clique problem on the corresponding graph classes.
0 citationsRead paper