Quantifying IIoT Sensor Node Criticality by Fusing its Data Criticality and Security Vulnerability
该研究通过结合数据重要性和网络安全漏洞，使用Dempster-Shafer理论评估IIoT传感器节点的关键性，提出了一种新的框架来解决工业物联网中的安全问题。
0 citationsRead paper
Selected work
Representative Papers
SDARE-Bench: Evaluating Large Language Models on Conversational Stigma Detection and Response in Dyadic and Group Dialogue
研究通过SDARE-Bench评估大语言模型在对话中检测和响应污名化问题的能力，揭示了模型在群体对话中的表现较差。
0 citationsRead paper
It's a matter of timescale: non-linear utility in successor features and multi-objective planning and learning
论文指出当前多目标强化学习方法无法处理不同时间尺度上的非线性效用问题，并通过实例说明该问题的重要性。
0 citationsRead paper
A Phased Development Framework Enabling Islanded Operation of Sustainable AI Data Centers With Onsite Grid-Following and Grid-Forming Energy Architectures
This study addresses engineering constraints in large-scale AI data center deployments, such as prolonged grid interconnection approval timelines and equipment delivery delays. To overcome these challenges, the authors propose a phased energy deployment architecture that integrates modular construction with a hybrid on-site energy system combining natural gas generation and grid-forming energy storage. This system enables islanded operation prior to full grid connection and facilitates seamless transition to grid-tied mode through a hybrid control strategy blending grid-forming and grid-following inverters. Electromagnetic transient simulations and modular design validation demonstrate that the hybrid system reliably supports high-power loads from early to mid-deployment stages and effectively manages islanding, reconnection, and recovery under grid disturbances. The approach significantly shortens construction timelines while enhancing power supply reliability and sustainability.
0 citationsRead paper
Filtering-out poor-quality images for data preparation
Traditional image denoising methods are constrained by specific noise types and fixed image dimensions, limiting their effectiveness in handling multi-source image degradation under complex real-world conditions. This work proposes an innovative data preprocessing strategy that abandons conventional denoising in favor of directly selecting high-quality images based on image quality assessment metrics and adaptive optimal thresholds, thereby constructing a high-fidelity training set suitable for deep learning. The approach requires no uniform image resizing and is compatible with diverse noise types and acquisition environments. Evaluated on traffic sign and general object recognition tasks, models trained with this method achieve average accuracies of 93.8% and 84.9%, respectively, significantly outperforming existing approaches and demonstrating strong efficacy and generalization capability for practical applications such as autonomous driving.
0 citationsRead paper
Recent publications
Latest Papers
Quantifying IIoT Sensor Node Criticality by Fusing its Data Criticality and Security Vulnerability
该研究通过结合数据重要性和网络安全漏洞，使用Dempster-Shafer理论评估IIoT传感器节点的关键性，提出了一种新的框架来解决工业物联网中的安全问题。
0 citationsRead paper
SDARE-Bench: Evaluating Large Language Models on Conversational Stigma Detection and Response in Dyadic and Group Dialogue
研究通过SDARE-Bench评估大语言模型在对话中检测和响应污名化问题的能力，揭示了模型在群体对话中的表现较差。
0 citationsRead paper
It's a matter of timescale: non-linear utility in successor features and multi-objective planning and learning
论文指出当前多目标强化学习方法无法处理不同时间尺度上的非线性效用问题，并通过实例说明该问题的重要性。
0 citationsRead paper
A Phased Development Framework Enabling Islanded Operation of Sustainable AI Data Centers With Onsite Grid-Following and Grid-Forming Energy Architectures
This study addresses engineering constraints in large-scale AI data center deployments, such as prolonged grid interconnection approval timelines and equipment delivery delays. To overcome these challenges, the authors propose a phased energy deployment architecture that integrates modular construction with a hybrid on-site energy system combining natural gas generation and grid-forming energy storage. This system enables islanded operation prior to full grid connection and facilitates seamless transition to grid-tied mode through a hybrid control strategy blending grid-forming and grid-following inverters. Electromagnetic transient simulations and modular design validation demonstrate that the hybrid system reliably supports high-power loads from early to mid-deployment stages and effectively manages islanding, reconnection, and recovery under grid disturbances. The approach significantly shortens construction timelines while enhancing power supply reliability and sustainability.
0 citationsRead paper
Filtering-out poor-quality images for data preparation
Traditional image denoising methods are constrained by specific noise types and fixed image dimensions, limiting their effectiveness in handling multi-source image degradation under complex real-world conditions. This work proposes an innovative data preprocessing strategy that abandons conventional denoising in favor of directly selecting high-quality images based on image quality assessment metrics and adaptive optimal thresholds, thereby constructing a high-fidelity training set suitable for deep learning. The approach requires no uniform image resizing and is compatible with diverse noise types and acquisition environments. Evaluated on traffic sign and general object recognition tasks, models trained with this method achieve average accuracies of 93.8% and 84.9%, respectively, significantly outperforming existing approaches and demonstrating strong efficacy and generalization capability for practical applications such as autonomous driving.
0 citationsRead paper