LUTSeg: A Longitudinal Multi-Expert Dataset for Ulcer Tissue Segmentation
为了解决慢性溃疡组织分割数据稀缺问题，通过构建LUTSeg数据集，并提出TiSage半监督分割框架来提高分割精度。
0 citationsRead paper
Selected work
Representative Papers
MARCIM-WG: A cyber wargame proposal based on math modeling applied in a naval scenario
This study addresses the critical vulnerability of maritime operations to cyber threats due to their heavy reliance on interconnected digital systems, emphasizing the urgent need to enhance decision-makers’ cyber situational awareness and incident response capabilities. To this end, the work proposes an innovative hybrid training system that integrates a physical wargaming board—featuring tokens and cards—with a simulation model driven by mathematical modeling–based computational adjudication. Structured crisis scenarios incorporate friction, resource constraints, and quantifiable consequences through high- and low-level design specifications. Evaluated via a tripartite validation framework (pessimistic, neutral, optimistic), the intervention group demonstrated a statistically significant 34.0-percentage-point improvement in cyber situational awareness, with particularly pronounced gains in comprehension-related competencies, thereby validating both the training efficacy and methodological novelty of the proposed approach.
0 citationsRead paper
Anticipating Adversary Behavior in DevSecOps Scenarios through Large Language Models
This work addresses the limitations of traditional DevSecOps practices, which often neglect proactive security integration and lack forward-looking models of attacker behavior, thereby struggling to defend cloud environments against sophisticated threats. To overcome these challenges, the paper introduces a novel automated approach that uniquely integrates large language models (LLMs) with Security Chaos Engineering (SCE). Specifically, LLMs are leveraged to generate attack-defense trees that simulate plausible attack paths, which in turn inform the design of SCE experiments. This methodology enables proactive prediction of unknown threats and facilitates the preemptive deployment of defensive strategies. Furthermore, it establishes a reproducible, LLM-driven security validation pipeline, significantly enhancing the proactive defense capabilities of DevSecOps teams.
0 citationsRead paper
Hierarchical Risk Parity for Portfolio Allocation in the Latin American NUAM Market
This study addresses the lack of suitable portfolio construction methodologies for the Latin American NUAM regional market (Chile, Colombia, Peru) under cross-border and highly volatile emerging-market conditions. Methodologically, it pioneers the application of Hierarchical Risk Parity (HRP) to this market, integrating hierarchical clustering with recursive bisection to circumvent covariance matrix inversion—thereby achieving robust, interpretable, and risk-balanced allocation. Empirical backtesting on daily returns of the 54 constituents of the MSCI NUAM Index demonstrates that HRP significantly reduces maximum drawdown and tracking error relative to equal-weighted and maximum-Sharpe-ratio portfolios, while maintaining competitive absolute returns and improving risk-adjusted performance. This work fills a critical gap in the literature by providing the first empirical validation of HRP in a regional emerging-market context and offers a replicable methodological framework for cross-national asset allocation in such environments.
0 citationsRead paper
Recent publications
Latest Papers
LUTSeg: A Longitudinal Multi-Expert Dataset for Ulcer Tissue Segmentation
为了解决慢性溃疡组织分割数据稀缺问题，通过构建LUTSeg数据集，并提出TiSage半监督分割框架来提高分割精度。
0 citationsRead paper
MARCIM-WG: A cyber wargame proposal based on math modeling applied in a naval scenario
This study addresses the critical vulnerability of maritime operations to cyber threats due to their heavy reliance on interconnected digital systems, emphasizing the urgent need to enhance decision-makers’ cyber situational awareness and incident response capabilities. To this end, the work proposes an innovative hybrid training system that integrates a physical wargaming board—featuring tokens and cards—with a simulation model driven by mathematical modeling–based computational adjudication. Structured crisis scenarios incorporate friction, resource constraints, and quantifiable consequences through high- and low-level design specifications. Evaluated via a tripartite validation framework (pessimistic, neutral, optimistic), the intervention group demonstrated a statistically significant 34.0-percentage-point improvement in cyber situational awareness, with particularly pronounced gains in comprehension-related competencies, thereby validating both the training efficacy and methodological novelty of the proposed approach.
0 citationsRead paper
Anticipating Adversary Behavior in DevSecOps Scenarios through Large Language Models
This work addresses the limitations of traditional DevSecOps practices, which often neglect proactive security integration and lack forward-looking models of attacker behavior, thereby struggling to defend cloud environments against sophisticated threats. To overcome these challenges, the paper introduces a novel automated approach that uniquely integrates large language models (LLMs) with Security Chaos Engineering (SCE). Specifically, LLMs are leveraged to generate attack-defense trees that simulate plausible attack paths, which in turn inform the design of SCE experiments. This methodology enables proactive prediction of unknown threats and facilitates the preemptive deployment of defensive strategies. Furthermore, it establishes a reproducible, LLM-driven security validation pipeline, significantly enhancing the proactive defense capabilities of DevSecOps teams.
0 citationsRead paper
Hierarchical Risk Parity for Portfolio Allocation in the Latin American NUAM Market
This study addresses the lack of suitable portfolio construction methodologies for the Latin American NUAM regional market (Chile, Colombia, Peru) under cross-border and highly volatile emerging-market conditions. Methodologically, it pioneers the application of Hierarchical Risk Parity (HRP) to this market, integrating hierarchical clustering with recursive bisection to circumvent covariance matrix inversion—thereby achieving robust, interpretable, and risk-balanced allocation. Empirical backtesting on daily returns of the 54 constituents of the MSCI NUAM Index demonstrates that HRP significantly reduces maximum drawdown and tracking error relative to equal-weighted and maximum-Sharpe-ratio portfolios, while maintaining competitive absolute returns and improving risk-adjusted performance. This work fills a critical gap in the literature by providing the first empirical validation of HRP in a regional emerging-market context and offers a replicable methodological framework for cross-national asset allocation in such environments.
0 citationsRead paper