Constrained Bayesian Optimization for Hierarchical Federated Learning in IoT Networks for Plant Disease Classification
本文提出了一种约束贝叶斯优化框架，用于在资源受限的物联网环境中高效配置分层联邦学习，以平衡能耗、执行时间和预测性能。
0 citationsRead paper
Selected work
Representative Papers
Earthquaker-AI: A Retrieval-Augmented Generation Framework with Rubric-Based Assessment for Primary School Earthquake Education
This study addresses the lack of integrated earthquake preparedness education for elementary students that effectively combines cognitive development, hands-on practice, and intelligent feedback. The work proposes an innovative system that integrates Lego WeDo2 robotics to simulate seismic scenarios with a retrieval-augmented generation (RAG)-based conversational AI. It introduces grade-adapted, multidimensional scoring rubrics to guide students in identifying, prioritizing, and articulating earthquake safety behaviors, thereby fostering self-regulated learning and calm crisis response. For the first time, RAG technology is coupled with an age-stratified, multidimensional assessment framework to advance learning from mechanical manipulation to cognitive reflection. The system strictly aligns with official safety guidelines to ensure factual accuracy, achieving high precision and low hallucination rates while significantly enhancing students’ earthquake knowledge, technological literacy, and early crisis-response capabilities.
0 citationsRead paper
On the Performance of DCF in Full Duplex WLANs with Hidden Terminals
This study addresses the limited throughput gains of full-duplex (FD) WLANs in hidden terminal scenarios, where performance remains constrained by the CSMA/CA-based Distributed Coordination Function (DCF). The paper presents the first systematic modeling and theoretical analysis of saturated throughput for both FD and half-duplex (HD) WLANs under DCF in such environments. By developing an accurate analytical model, the work quantifies the throughput improvement achievable with FD operation and reveals that, under typical parameter settings, FD yields only a modest 10%–20% throughput gain over HD. These findings provide critical theoretical insights into the practical efficacy of FD in existing MAC-layer protocols and offer valuable guidance for future protocol design.
0 citationsRead paper
Throughput and Delay Performance of Slotted Aloha in SmartBANs under Saturation Conditions
This study addresses the throughput and end-to-end delay performance of the slotted ALOHA protocol under saturated conditions as specified in the ETSI SmartBAN standard. The work presents the first accurate two-dimensional discrete-time Markov chain (DTMC) analytical model that fully captures the protocol’s dynamic behavior. By integrating theoretical modeling with system-level simulations, the proposed framework effectively elucidates the operational mechanisms of the protocol and enables highly precise predictions of both saturation throughput and average packet delay. The close agreement between analytical results and simulation outcomes validates the model’s accuracy, establishing it as a reliable theoretical tool for performance evaluation and parameter optimization of SmartBAN protocols.
0 citationsRead paper
Explainability for Fault Detection System in Chemical Processes
This study addresses the lack of interpretability in fault detection for chemical processes by proposing a diagnostic framework that integrates a high-accuracy LSTM classifier with explainable artificial intelligence (XAI) techniques. It presents the first systematic comparison of Integrated Gradients (IG) and SHAP in interpreting faults within complex nonlinear chemical processes, specifically using the Tennessee Eastman benchmark. The evaluation demonstrates that SHAP more accurately identifies root causes and key contributing variables. The resulting model-agnostic explainability framework not only effectively localizes faulty subsystems but also exhibits strong generalization across diverse operational scenarios. This approach provides reliable and transparent decision support for safety monitoring in industrial processes, enhancing both trustworthiness and practical utility in real-world applications.
0 citationsRead paper
Recent publications
Latest Papers
Constrained Bayesian Optimization for Hierarchical Federated Learning in IoT Networks for Plant Disease Classification
本文提出了一种约束贝叶斯优化框架，用于在资源受限的物联网环境中高效配置分层联邦学习，以平衡能耗、执行时间和预测性能。
0 citationsRead paper
Earthquaker-AI: A Retrieval-Augmented Generation Framework with Rubric-Based Assessment for Primary School Earthquake Education
This study addresses the lack of integrated earthquake preparedness education for elementary students that effectively combines cognitive development, hands-on practice, and intelligent feedback. The work proposes an innovative system that integrates Lego WeDo2 robotics to simulate seismic scenarios with a retrieval-augmented generation (RAG)-based conversational AI. It introduces grade-adapted, multidimensional scoring rubrics to guide students in identifying, prioritizing, and articulating earthquake safety behaviors, thereby fostering self-regulated learning and calm crisis response. For the first time, RAG technology is coupled with an age-stratified, multidimensional assessment framework to advance learning from mechanical manipulation to cognitive reflection. The system strictly aligns with official safety guidelines to ensure factual accuracy, achieving high precision and low hallucination rates while significantly enhancing students’ earthquake knowledge, technological literacy, and early crisis-response capabilities.
0 citationsRead paper
On the Performance of DCF in Full Duplex WLANs with Hidden Terminals
This study addresses the limited throughput gains of full-duplex (FD) WLANs in hidden terminal scenarios, where performance remains constrained by the CSMA/CA-based Distributed Coordination Function (DCF). The paper presents the first systematic modeling and theoretical analysis of saturated throughput for both FD and half-duplex (HD) WLANs under DCF in such environments. By developing an accurate analytical model, the work quantifies the throughput improvement achievable with FD operation and reveals that, under typical parameter settings, FD yields only a modest 10%–20% throughput gain over HD. These findings provide critical theoretical insights into the practical efficacy of FD in existing MAC-layer protocols and offer valuable guidance for future protocol design.
0 citationsRead paper
Throughput and Delay Performance of Slotted Aloha in SmartBANs under Saturation Conditions
This study addresses the throughput and end-to-end delay performance of the slotted ALOHA protocol under saturated conditions as specified in the ETSI SmartBAN standard. The work presents the first accurate two-dimensional discrete-time Markov chain (DTMC) analytical model that fully captures the protocol’s dynamic behavior. By integrating theoretical modeling with system-level simulations, the proposed framework effectively elucidates the operational mechanisms of the protocol and enables highly precise predictions of both saturation throughput and average packet delay. The close agreement between analytical results and simulation outcomes validates the model’s accuracy, establishing it as a reliable theoretical tool for performance evaluation and parameter optimization of SmartBAN protocols.
0 citationsRead paper
Explainability for Fault Detection System in Chemical Processes
This study addresses the lack of interpretability in fault detection for chemical processes by proposing a diagnostic framework that integrates a high-accuracy LSTM classifier with explainable artificial intelligence (XAI) techniques. It presents the first systematic comparison of Integrated Gradients (IG) and SHAP in interpreting faults within complex nonlinear chemical processes, specifically using the Tennessee Eastman benchmark. The evaluation demonstrates that SHAP more accurately identifies root causes and key contributing variables. The resulting model-agnostic explainability framework not only effectively localizes faulty subsystems but also exhibits strong generalization across diverse operational scenarios. This approach provides reliable and transparent decision support for safety monitoring in industrial processes, enhancing both trustworthiness and practical utility in real-world applications.
0 citationsRead paper