The rapid and accurate assessment of structural damage following an earthquake is crucial for effective emergency response and post-disaster recovery. Traditional manual inspection methods are often slow, labor-intens...The rapid and accurate assessment of structural damage following an earthquake is crucial for effective emergency response and post-disaster recovery. Traditional manual inspection methods are often slow, labor-intensive, and prone to human error. To address these challenges, this study proposes STPEIC (Swin Transformer-based Framework for Interpretable Post-Earthquake Structural Classification), an automated deep learning framework designed for analyzing post-earthquake images. STPEIC performs two key tasks: structural components classification and damage level classification. By leveraging the hierarchical attention mechanisms of the Swin Transformer (Shifted Window Transformer), the model achieves 85.4% accuracy in structural component classification and 85.1% accuracy in damage level classification. To enhance model interpretability, visual explanation heatmaps are incorporated, highlighting semantically relevant regions that the model uses for decision-making. These heatmaps closely align with real-world structural and damage features, confirming that STPEIC learns meaningful representations rather than relying on spurious correlations. Additionally, a graphical user interface (GUI) has been developed to streamline image input, classification, and interpretability visualization, improving the practical usability of the system. Overall, STPEIC provides a reliable, interpretable, and user-friendly solution for rapid post-earthquake structural evaluation.展开更多
Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable resul...Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable result of knowledge status.Markov chain-based knowledge tracing(MCKT)models,such as Bayesian Knowledge Tracing,can track knowledge concept mastery probability over time.However,as the number of tracked knowledge concepts increases,the time complexity of MCKT predicting student performance increases exponentially(also called explaining away problem).When the number of tracked knowledge concepts is large,we cannot utilize MCKT to track knowledge concept mastery probability over time.In addition,the existing MCKT models only consider the relationship between students’knowledge status and problems when modeling students’responses but ignore the relationship between knowledge concepts in the same problem.To address these challenges,we propose an inTerpretable pRobAbilistiC gEnerative moDel(TRACED),which can track students’numerous knowledge concepts mastery probabilities over time.To solve explain away problem,we design long and short-term memory(LSTM)-based networks to approximate the posterior distribution,predict students’future performance,and propose a heuristic algorithm to train LSTMs and probabilistic graphical model jointly.To better model students’exercise responses,we proposed a logarithmic linear model with three interactive strategies,which models students’exercise responses by considering the relationship among students’knowledge status,knowledge concept,and problems.We conduct experiments with four real-world datasets in three knowledge-driven tasks.The experimental results show that TRACED outperforms existing knowledge tracing methods in predicting students’future performance and can learn the relationship among students,knowledge concepts,and problems from students’exercise sequences.We also conduct several case studies.The case studies show that TRACED exhibits excellent interpretability and thus has the potential for personalized automatic feedback in the real-world educational environment.展开更多
This paper studies the renewable power forecasting task with a more advanced formulation,the probabilistic forecasts of day-ahead power generation sequences of multiple renewable power plants without breaching the pri...This paper studies the renewable power forecasting task with a more advanced formulation,the probabilistic forecasts of day-ahead power generation sequences of multiple renewable power plants without breaching the privacy of data in each plant.To realize such a task,an advanced domain-invariant feature learning embedded federated learning(DIFL)framework is proposed to coordinate the development of a system of deep networkbased models serving as multiple clients and one server.In DIFL,each client,which serves each local renew-able power plant,maps its raw data input into latent features via a local feature extractor and generates power output sequence probabilistic forecasts via a locally hosted forecasting model.The cloud-hosted server first aggregates the knowledge from models of clients and next dispatches the aggregated model back to each client for facilitating each local feature extractor to identify domain-invariant features via interacting with a server-side discriminator.Therefore,only desensitized data,such as parameters of the models,are allowed to be transmitted among end users for preserving local data privacy of power plants.To verify the advantages of the DIFL,a preliminary exploration of its theoretical property is first conducted.Next,computational studies are performed to benchmark the DIFL against famous baselines based on datasets collected from commercial renewable power plants.Results further confirm that,in terms of the averaged performance,the DIFL consistently realizes im-provements against all benchmarks based on both real wind farm and solar power plant datasets.展开更多
基金support from General Project Supported by Natural Science Basic Research Plan in Shaanxi Province of China(2025JC-YBMS-443)Fundamental Research Funds for the Central Universities,CHU(300102213209)+1 种基金Research Funds for the Interdisciplinary Projects,CHU(300104240915)National Natural Science Foundation of China(Grant No.52361135806).
文摘The rapid and accurate assessment of structural damage following an earthquake is crucial for effective emergency response and post-disaster recovery. Traditional manual inspection methods are often slow, labor-intensive, and prone to human error. To address these challenges, this study proposes STPEIC (Swin Transformer-based Framework for Interpretable Post-Earthquake Structural Classification), an automated deep learning framework designed for analyzing post-earthquake images. STPEIC performs two key tasks: structural components classification and damage level classification. By leveraging the hierarchical attention mechanisms of the Swin Transformer (Shifted Window Transformer), the model achieves 85.4% accuracy in structural component classification and 85.1% accuracy in damage level classification. To enhance model interpretability, visual explanation heatmaps are incorporated, highlighting semantically relevant regions that the model uses for decision-making. These heatmaps closely align with real-world structural and damage features, confirming that STPEIC learns meaningful representations rather than relying on spurious correlations. Additionally, a graphical user interface (GUI) has been developed to streamline image input, classification, and interpretability visualization, improving the practical usability of the system. Overall, STPEIC provides a reliable, interpretable, and user-friendly solution for rapid post-earthquake structural evaluation.
基金supported by the National Natural Science Foundation of China(Grant Nos.62272093,62137001,U1811261,and 61902055).
文摘Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable result of knowledge status.Markov chain-based knowledge tracing(MCKT)models,such as Bayesian Knowledge Tracing,can track knowledge concept mastery probability over time.However,as the number of tracked knowledge concepts increases,the time complexity of MCKT predicting student performance increases exponentially(also called explaining away problem).When the number of tracked knowledge concepts is large,we cannot utilize MCKT to track knowledge concept mastery probability over time.In addition,the existing MCKT models only consider the relationship between students’knowledge status and problems when modeling students’responses but ignore the relationship between knowledge concepts in the same problem.To address these challenges,we propose an inTerpretable pRobAbilistiC gEnerative moDel(TRACED),which can track students’numerous knowledge concepts mastery probabilities over time.To solve explain away problem,we design long and short-term memory(LSTM)-based networks to approximate the posterior distribution,predict students’future performance,and propose a heuristic algorithm to train LSTMs and probabilistic graphical model jointly.To better model students’exercise responses,we proposed a logarithmic linear model with three interactive strategies,which models students’exercise responses by considering the relationship among students’knowledge status,knowledge concept,and problems.We conduct experiments with four real-world datasets in three knowledge-driven tasks.The experimental results show that TRACED outperforms existing knowledge tracing methods in predicting students’future performance and can learn the relationship among students,knowledge concepts,and problems from students’exercise sequences.We also conduct several case studies.The case studies show that TRACED exhibits excellent interpretability and thus has the potential for personalized automatic feedback in the real-world educational environment.
基金supported in part by the Hong Kong RGC General Research Fund Project with No.11213124in part by Hong Kong ITC Innovation and Technology Fund Project with No.ITS/034/22MS+2 种基金in part by in part by Guangdong Provincial Basic and Applied Basic Research-Offshore Wind Power Joint Fund Project under Grant 2022A1515240066in part by Guangdong Province Technological Project with No.2023A0505030014in part by the Shenzhen-Hong Kong-Macao Science&Technology Category C Project with No.SGDX20220530111205037.
文摘This paper studies the renewable power forecasting task with a more advanced formulation,the probabilistic forecasts of day-ahead power generation sequences of multiple renewable power plants without breaching the privacy of data in each plant.To realize such a task,an advanced domain-invariant feature learning embedded federated learning(DIFL)framework is proposed to coordinate the development of a system of deep networkbased models serving as multiple clients and one server.In DIFL,each client,which serves each local renew-able power plant,maps its raw data input into latent features via a local feature extractor and generates power output sequence probabilistic forecasts via a locally hosted forecasting model.The cloud-hosted server first aggregates the knowledge from models of clients and next dispatches the aggregated model back to each client for facilitating each local feature extractor to identify domain-invariant features via interacting with a server-side discriminator.Therefore,only desensitized data,such as parameters of the models,are allowed to be transmitted among end users for preserving local data privacy of power plants.To verify the advantages of the DIFL,a preliminary exploration of its theoretical property is first conducted.Next,computational studies are performed to benchmark the DIFL against famous baselines based on datasets collected from commercial renewable power plants.Results further confirm that,in terms of the averaged performance,the DIFL consistently realizes im-provements against all benchmarks based on both real wind farm and solar power plant datasets.
