Repeatability Characterisation and Error Budget of a Consumer Structured-Light Scanner for 3-D Wound Geometry:A Rigid-Phantom Study
研究使用消费级结构光扫描仪测量三维伤口几何形状的误差，通过重复扫描固定模型来评估其可重复性及误差来源。
0 citationsRead paper
Selected work
Representative Papers
StealthBench: Measuring Operational Stealth in Autonomous Offensive-Security Agents
This work addresses the critical issue that current autonomous offensive–defensive agents often compromise operational security (OPSEC) by exhibiting non-covert behaviors, leading to exposure. To systematically evaluate agent stealthiness, we introduce StealthBench—the first benchmark specifically designed for assessing OPSEC compliance, encompassing six core OPSEC dimensions and 14 containerized scenarios derived from real-world red-team engagements and bug bounty reports. We propose composite metrics including Safe Success Rate, Stealth@Solve, and Reckless Solve Rate, and employ a three-model LLM adjudication framework with majority voting to determine behavioral compliance. Empirical evaluation reveals that even state-of-the-art models achieve safe success rates below 54%, underscoring the pervasiveness of OPSEC violations. The benchmark, evaluation framework, and an interactive leaderboard are publicly released.
0 citationsRead paper
How Often Should a Recommender Call an LLM? Value-Weighted Routing, Monitoring, and Seasonal Robustness
This work addresses a critical limitation in traditional recommendation systems that route queries between lightweight heuristics and large language models solely based on task difficulty, ignoring disparities in error cost and business value. To overcome this, the authors propose a value-weighted routing mechanism that makes unsupervised routing decisions by jointly estimating task difficulty and item-level business value within a fully synthetic retail assortment simulation environment. The framework incorporates decision logging and monitoring modules to uncover category-level biases obscured by aggregate metrics. Through slow-path budget control and seasonal parameter tuning, the system achieves a 60% recall rate on high-value items while improving overall accuracy from 94.3% to 98.3%, demonstrating enhanced robustness under simulated Black Friday traffic surges.
0 citationsRead paper
FRAMEWORK FOR INTEGRATING ZERO TRUST IN CLOUD-BASED ENDPOINT SECURITY FOR CRITICAL INFRASTRUCTURE
This work proposes a novel endpoint security framework tailored for critical infrastructure operating in cloud environments, where increasingly sophisticated endpoint threats challenge conventional security models and zero trust architecture (ZTA) remains underexplored for endpoint management. The framework systematically integrates ZTA principles into endpoint security by enforcing continuous authentication, least-privilege access control, and cloud-native security mechanisms to dynamically validate every access request. This approach significantly reduces the attack surface, enhances endpoint protection capabilities, and strengthens compliance with regulatory requirements. By embedding zero trust deeply into the endpoint layer of critical infrastructure cloud systems, the proposed framework addresses a critical gap in current zero trust practices and offers a robust foundation for securing high-value assets against evolving threats.
0 citationsRead paper
Recent publications
Latest Papers
Repeatability Characterisation and Error Budget of a Consumer Structured-Light Scanner for 3-D Wound Geometry:A Rigid-Phantom Study
研究使用消费级结构光扫描仪测量三维伤口几何形状的误差，通过重复扫描固定模型来评估其可重复性及误差来源。
0 citationsRead paper
StealthBench: Measuring Operational Stealth in Autonomous Offensive-Security Agents
This work addresses the critical issue that current autonomous offensive–defensive agents often compromise operational security (OPSEC) by exhibiting non-covert behaviors, leading to exposure. To systematically evaluate agent stealthiness, we introduce StealthBench—the first benchmark specifically designed for assessing OPSEC compliance, encompassing six core OPSEC dimensions and 14 containerized scenarios derived from real-world red-team engagements and bug bounty reports. We propose composite metrics including Safe Success Rate, Stealth@Solve, and Reckless Solve Rate, and employ a three-model LLM adjudication framework with majority voting to determine behavioral compliance. Empirical evaluation reveals that even state-of-the-art models achieve safe success rates below 54%, underscoring the pervasiveness of OPSEC violations. The benchmark, evaluation framework, and an interactive leaderboard are publicly released.
0 citationsRead paper
How Often Should a Recommender Call an LLM? Value-Weighted Routing, Monitoring, and Seasonal Robustness
This work addresses a critical limitation in traditional recommendation systems that route queries between lightweight heuristics and large language models solely based on task difficulty, ignoring disparities in error cost and business value. To overcome this, the authors propose a value-weighted routing mechanism that makes unsupervised routing decisions by jointly estimating task difficulty and item-level business value within a fully synthetic retail assortment simulation environment. The framework incorporates decision logging and monitoring modules to uncover category-level biases obscured by aggregate metrics. Through slow-path budget control and seasonal parameter tuning, the system achieves a 60% recall rate on high-value items while improving overall accuracy from 94.3% to 98.3%, demonstrating enhanced robustness under simulated Black Friday traffic surges.
0 citationsRead paper
FRAMEWORK FOR INTEGRATING ZERO TRUST IN CLOUD-BASED ENDPOINT SECURITY FOR CRITICAL INFRASTRUCTURE
This work proposes a novel endpoint security framework tailored for critical infrastructure operating in cloud environments, where increasingly sophisticated endpoint threats challenge conventional security models and zero trust architecture (ZTA) remains underexplored for endpoint management. The framework systematically integrates ZTA principles into endpoint security by enforcing continuous authentication, least-privilege access control, and cloud-native security mechanisms to dynamically validate every access request. This approach significantly reduces the attack surface, enhances endpoint protection capabilities, and strengthens compliance with regulatory requirements. By embedding zero trust deeply into the endpoint layer of critical infrastructure cloud systems, the proposed framework addresses a critical gap in current zero trust practices and offers a robust foundation for securing high-value assets against evolving threats.
0 citationsRead paper