Predictive Business Process Monitoring(PBPM)is a significant research area in Business Process Management(BPM)aimed at accurately forecasting future behavioral events.At present,deep learning methods are widely cited ...Predictive Business Process Monitoring(PBPM)is a significant research area in Business Process Management(BPM)aimed at accurately forecasting future behavioral events.At present,deep learning methods are widely cited in PBPM research,but no method has been effective in fusing data information into the control flow for multi-perspective process prediction.Therefore,this paper proposes a process prediction method based on the hierarchical BERT and multi-perspective data fusion.Firstly,the first layer BERT network learns the correlations between different category attribute data.Then,the attribute data is integrated into a weighted event-level feature vector and input into the second layer BERT network to learn the impact and priority relationship of each event on future predicted events.Next,the multi-head attention mechanism within the framework is visualized for analysis,helping to understand the decision-making logic of the framework and providing visual predictions.Finally,experimental results show that the predictive accuracy of the framework surpasses the current state-of-the-art research methods and significantly enhances the predictive performance of BPM.展开更多
With the wide application of location-based social networks(LBSNs),personalized point of interest(POI)recommendation becomes popular,especially in the commercial field.Unfortunately,it is challenging to accurately rec...With the wide application of location-based social networks(LBSNs),personalized point of interest(POI)recommendation becomes popular,especially in the commercial field.Unfortunately,it is challenging to accurately recommend POIs to users because the user-POI matrix is extremely sparse.In addition,a user's check-in activities are affected by many influential factors.However,most of existing studies capture only few influential factors.It is hard for them to be extended to incorporate other heterogeneous information in a unified way.To address these problems,we propose a meta-path-based deep representation learning(MPDRL)model for personalized POI recommendation.In this model,we design eight types of meta-paths to fully utilize the rich heterogeneous information in LBSNs for the representations of users and POIs,and deeply mine the correlations between users and POIs.To further improve the recommendation performance,we design an attention-based long short-term memory(LSTM)network to learn the importance of different influential factors on a user's specific check-in activity.To verify the effectiveness of our proposed method,we conduct extensive experiments on a real-world dataset,Foursquare.Experimental results show that the MPDRL model improves at least 16.97%and 23.55%over all comparison methods in terms of the metric Precision@N(Pre@N)and Recall@N(Rec@N)respectively.展开更多
Continuous and accurate blood pressure(BP)monitoring is critical for personalized hypertension management.However,most existing methods focus on absolute BP estimation,with limited attention to BP changes.To address t...Continuous and accurate blood pressure(BP)monitoring is critical for personalized hypertension management.However,most existing methods focus on absolute BP estimation,with limited attention to BP changes.To address this limitation,we propose a novel framework named Multi-Perspective Differential Feature Space(MDFSBP)for cuffless BP estimation using photoplethysmography(PPG)signals.MDFSBP extracts three perspective differential features:time-based and points-of-interest features,frequency-domain features,and physiological statistical features.Then,an adaptive Multi-Perspective Differential Feature Mapping Module(MDFMM)integrates reconstruction regularization,trend-aware classification,and self-weighted contrastive learning to enhance feature representation and strengthen the association between features and BP changes.Finally,an AutoML-based regression pipeline automates model optimization,improving predictive accuracy and deployment efficiency.To better test the model's capability in capturing BP changes,we introduce a novel abnormality-aware classification metric.We demonstrate BP estimation performance over state-of-the-art(SOTA)methods on both the Mindray and MIMIC datasets.On the Mindray dataset,the model achieves a regression error of 0.17±5.17 mmHg for SBP and 0.05±3.29 mmHg for DBP,with classification accuracy and F1-score reaching 85.25%and 87.50%,respectively.On the MIMIC dataset,it achieves−0.09±5.70 mmHg for SBP and 0.12±4.27 mmHg for DBP,with the classification accuracy and F1-score of 72.84%and 70.66%,respectively.These results highlight the effectiveness,robustness,and generalizability of the proposed frame-work for non-invasive,real-time,and continuous BP monitoring in both clinical and wearable healthcare systems.展开更多
A novel 3 D digital image correlation(DIC) system based on a single three charge-couple device(3 CCD) color camera is proposed in this paper. Images from three different perspectives are captured by a 3 CCD camera usi...A novel 3 D digital image correlation(DIC) system based on a single three charge-couple device(3 CCD) color camera is proposed in this paper. Images from three different perspectives are captured by a 3 CCD camera using a reflective-based pseudo-vision system. These images are then separated by the different CCD channels, and the correlation algorithm for the multi-camera DIC system is adopted to evaluate the images. Compared to the conventional multi-camera DIC system, the proposed system is much more compact. In addition, the proposed system has no loss of spatial resolution, compares to the traditional single camera DIC system. The complex surface measurement ability and the measurement accuracy is significantly improved through the use of the multi-camera DIC algorithm. The principle of the proposed system is described in detail as well as the experimental setup. A series of validation tests are performed, and the results are verified with the commercial 3 D-DIC system.展开更多
Information networks that can be extracted from many domains are widely studied recently. Different functions for mining these networks are proposed and developed, such as ranking, community detection, and link predic...Information networks that can be extracted from many domains are widely studied recently. Different functions for mining these networks are proposed and developed, such as ranking, community detection, and link prediction. Most existing network studies are on homogeneous networks, where nodes and links are assumed from one single type. In reality, however, heterogeneous information networks can better model the real-world systems, which are typically semi-structured and typed, following a network schema. In order to mine these heterogeneous information networks directly, we propose to explore the meta structure of the information network, i.e., the network schema. The concepts of meta-paths are proposed to systematically capture numerous semantic relationships across multiple types of objects, which are defined as a path over the graph of network schema. Meta-paths can provide guidance for search and mining of the network and help analyze and understand the semantic meaning of the objects and relations in the network. Under this framework, similarity search and other mining tasks such as relationship prediction and clustering can be addressed by systematic exploration of the network meta structure. Moreover, with user＇s guidance or feedback, we can select the best meta-path or their weighted combination for a specific mining task.展开更多
Commonsense question answering(CQA)requires understanding and reasoning over QA context and related commonsense knowledge,such as a structured Knowledge Graph(KG).Existing studies combine language models and graph neu...Commonsense question answering(CQA)requires understanding and reasoning over QA context and related commonsense knowledge,such as a structured Knowledge Graph(KG).Existing studies combine language models and graph neural networks to model inference.However,traditional knowledge graph are mostly concept-based,ignoring direct path evidence necessary for accurate reasoning.In this paper,we propose MRGNN(Meta-path Reasoning Graph Neural Network),a novel model that comprehensively captures sequential semantic information from concepts and paths.In MRGNN,meta-paths are introduced as direct inference evidence and an original graph neural network is adopted to aggregate features from both concepts and paths simultaneously.We conduct sufficient experiments on the CommonsenceQA and OpenBookQA datasets,showing the effectiveness of MRGNN.Also,we conduct further ablation experiments and explain the reasoning behavior through the case study.展开更多
Drug-drug interaction(DDI)event prediction is a challenging problem,and accurate prediction of DDI events is critical to patient health and new drug development.Recently,many machine learning-based techniques have bee...Drug-drug interaction(DDI)event prediction is a challenging problem,and accurate prediction of DDI events is critical to patient health and new drug development.Recently,many machine learning-based techniques have been proposed for predicting DDI events.However,most of the existing methods do not effectively integrate the multidimensional features of drugs and provide poor mitigation of noise to get effective feature information.To address these limitations,we propose a DDI-Transform neural network framework for DDI event prediction.In DDI-Transform,we design a drug structure information feature extraction module and a drug bind-protein feature extraction module to obtain multidimensional feature information.A stack of DDI-Transform layers in the DDI-Transform network module are then used for adaptive learning,thus adaptively selecting the effective feature information for prediction.The results show that DDI-Transform can accurately predict DDI events and outperform the state-of-the-art models.Results on different scale datasets confirm the robustness of the method.展开更多
Most entity ranking research aims to retrieve a ranked list of entities from a Web corpus given a user query. The rank order of entities is determined by the relevance between the query and contexts of entities. Howev...Most entity ranking research aims to retrieve a ranked list of entities from a Web corpus given a user query. The rank order of entities is determined by the relevance between the query and contexts of entities. However, entities can be ranked directly based on their relative importance in a document collection, independent of any queries. In this paper, we introduce an entity ranking algorithm named NERank＋. Given a document collection, NERank＋ first constructs a graph model called Topical Tripartite Graph, consisting of document, topic and entity nodes. We design separate ranking functions to compute the prior ranks of entities and topics, respectively. A meta-path constrained random walk algorithm is proposed to propagate prior entity and topic ranks based on the graph model. We evaluate NERank＋ over real-life datasets and compare it with baselines. Experimental results illustrate the effectiveness of our approach.展开更多
The present study proposed a shaped sweeping jet(SJ)that possesses the merits of both SJ and shaped hole,which demonstrates significantly improved cooling effectiveness and anti-deposition performance.Compared to a cl...The present study proposed a shaped sweeping jet(SJ)that possesses the merits of both SJ and shaped hole,which demonstrates significantly improved cooling effectiveness and anti-deposition performance.Compared to a classical 777 shaped hole,the shaped SJ exhibits a maximum enhancement of 70%in cooling effectiveness and a maximum reduction of 28%in particle deposition height,respectively.Owing to the periodic oscillation of coolant jet and higher streamwise jet momentum,the shaped SJ can provide much wider coolant coverage and therefore sweep the adhesive particle away from the wall.This study is the first attempt to reconcile the performance of film cooling and particle anti-deposition simultaneously,which offers a promising design concept for future engine cooling.展开更多
In this paper, we proposed a new kind of mark points coded by color and a new quasi-ellipse detector on pixel level. This method is especially applicable to three- dimensional （3D） head panoramic reconstruction. Ima...In this paper, we proposed a new kind of mark points coded by color and a new quasi-ellipse detector on pixel level. This method is especially applicable to three- dimensional （3D） head panoramic reconstruction. Images of adjacent perspectives can be stitched by matching pasted color-coded mark points in overlap area to calculate the transformation matrix. This paper focuses on how the color-coded mark points work and how to detect and match corresponding points from different perspectives. Tests are performed to show the efficiency and accuracy of this method based on the original data obtained by structured light projection.展开更多
基金Supported by the National Natural Science Foundation,China(No.61402011)the Open Project Program of the Key Laboratory of Embedded System and Service Computing of Ministry of Education(No.ESSCKF2021-05).
文摘Predictive Business Process Monitoring(PBPM)is a significant research area in Business Process Management(BPM)aimed at accurately forecasting future behavioral events.At present,deep learning methods are widely cited in PBPM research,but no method has been effective in fusing data information into the control flow for multi-perspective process prediction.Therefore,this paper proposes a process prediction method based on the hierarchical BERT and multi-perspective data fusion.Firstly,the first layer BERT network learns the correlations between different category attribute data.Then,the attribute data is integrated into a weighted event-level feature vector and input into the second layer BERT network to learn the impact and priority relationship of each event on future predicted events.Next,the multi-head attention mechanism within the framework is visualized for analysis,helping to understand the decision-making logic of the framework and providing visual predictions.Finally,experimental results show that the predictive accuracy of the framework surpasses the current state-of-the-art research methods and significantly enhances the predictive performance of BPM.
基金National Natural Science Foundation of China(No.61972080)Shanghai Rising-Star Program,China(No.19QA1400300)。
文摘With the wide application of location-based social networks(LBSNs),personalized point of interest(POI)recommendation becomes popular,especially in the commercial field.Unfortunately,it is challenging to accurately recommend POIs to users because the user-POI matrix is extremely sparse.In addition,a user's check-in activities are affected by many influential factors.However,most of existing studies capture only few influential factors.It is hard for them to be extended to incorporate other heterogeneous information in a unified way.To address these problems,we propose a meta-path-based deep representation learning(MPDRL)model for personalized POI recommendation.In this model,we design eight types of meta-paths to fully utilize the rich heterogeneous information in LBSNs for the representations of users and POIs,and deeply mine the correlations between users and POIs.To further improve the recommendation performance,we design an attention-based long short-term memory(LSTM)network to learn the importance of different influential factors on a user's specific check-in activity.To verify the effectiveness of our proposed method,we conduct extensive experiments on a real-world dataset,Foursquare.Experimental results show that the MPDRL model improves at least 16.97%and 23.55%over all comparison methods in terms of the metric Precision@N(Pre@N)and Recall@N(Rec@N)respectively.
基金the Capital Medical Technol-ogy Development Project(2024-IG-4032)National Key Research and Development Program of China(No.2023YFC3603500)+1 种基金the National Natural Science Foundation of China(No.62271023)the Beijing Natural Science Foundation(No.7242269).
文摘Continuous and accurate blood pressure(BP)monitoring is critical for personalized hypertension management.However,most existing methods focus on absolute BP estimation,with limited attention to BP changes.To address this limitation,we propose a novel framework named Multi-Perspective Differential Feature Space(MDFSBP)for cuffless BP estimation using photoplethysmography(PPG)signals.MDFSBP extracts three perspective differential features:time-based and points-of-interest features,frequency-domain features,and physiological statistical features.Then,an adaptive Multi-Perspective Differential Feature Mapping Module(MDFMM)integrates reconstruction regularization,trend-aware classification,and self-weighted contrastive learning to enhance feature representation and strengthen the association between features and BP changes.Finally,an AutoML-based regression pipeline automates model optimization,improving predictive accuracy and deployment efficiency.To better test the model's capability in capturing BP changes,we introduce a novel abnormality-aware classification metric.We demonstrate BP estimation performance over state-of-the-art(SOTA)methods on both the Mindray and MIMIC datasets.On the Mindray dataset,the model achieves a regression error of 0.17±5.17 mmHg for SBP and 0.05±3.29 mmHg for DBP,with classification accuracy and F1-score reaching 85.25%and 87.50%,respectively.On the MIMIC dataset,it achieves−0.09±5.70 mmHg for SBP and 0.12±4.27 mmHg for DBP,with the classification accuracy and F1-score of 72.84%and 70.66%,respectively.These results highlight the effectiveness,robustness,and generalizability of the proposed frame-work for non-invasive,real-time,and continuous BP monitoring in both clinical and wearable healthcare systems.
基金supported by the National Natural Science Foundation of China(Grant Nos.51375136&11672045)
文摘A novel 3 D digital image correlation(DIC) system based on a single three charge-couple device(3 CCD) color camera is proposed in this paper. Images from three different perspectives are captured by a 3 CCD camera using a reflective-based pseudo-vision system. These images are then separated by the different CCD channels, and the correlation algorithm for the multi-camera DIC system is adopted to evaluate the images. Compared to the conventional multi-camera DIC system, the proposed system is much more compact. In addition, the proposed system has no loss of spatial resolution, compares to the traditional single camera DIC system. The complex surface measurement ability and the measurement accuracy is significantly improved through the use of the multi-camera DIC algorithm. The principle of the proposed system is described in detail as well as the experimental setup. A series of validation tests are performed, and the results are verified with the commercial 3 D-DIC system.
基金supported in part by the U.S.Army Research Laboratory under Cooperative Agreement No.W911NF-09-2-0053(NS-CTA),NSF ⅡS-0905215,CNS-09-31975MIAS,a DHS-IDS Center for Multimodal Information Access and Synthesis at UIUC
文摘Information networks that can be extracted from many domains are widely studied recently. Different functions for mining these networks are proposed and developed, such as ranking, community detection, and link prediction. Most existing network studies are on homogeneous networks, where nodes and links are assumed from one single type. In reality, however, heterogeneous information networks can better model the real-world systems, which are typically semi-structured and typed, following a network schema. In order to mine these heterogeneous information networks directly, we propose to explore the meta structure of the information network, i.e., the network schema. The concepts of meta-paths are proposed to systematically capture numerous semantic relationships across multiple types of objects, which are defined as a path over the graph of network schema. Meta-paths can provide guidance for search and mining of the network and help analyze and understand the semantic meaning of the objects and relations in the network. Under this framework, similarity search and other mining tasks such as relationship prediction and clustering can be addressed by systematic exploration of the network meta structure. Moreover, with user＇s guidance or feedback, we can select the best meta-path or their weighted combination for a specific mining task.
基金supported by the Key Research and Development Program of Hubei Province(2020BAB017)the Scientific Research Center Program of National Language Commission(ZDI135-135)the Fundamental Research Funds for the Central Universities(KJ02502022-0155,CCNU22XJ037).
文摘Commonsense question answering(CQA)requires understanding and reasoning over QA context and related commonsense knowledge,such as a structured Knowledge Graph(KG).Existing studies combine language models and graph neural networks to model inference.However,traditional knowledge graph are mostly concept-based,ignoring direct path evidence necessary for accurate reasoning.In this paper,we propose MRGNN(Meta-path Reasoning Graph Neural Network),a novel model that comprehensively captures sequential semantic information from concepts and paths.In MRGNN,meta-paths are introduced as direct inference evidence and an original graph neural network is adopted to aggregate features from both concepts and paths simultaneously.We conduct sufficient experiments on the CommonsenceQA and OpenBookQA datasets,showing the effectiveness of MRGNN.Also,we conduct further ablation experiments and explain the reasoning behavior through the case study.
文摘Drug-drug interaction(DDI)event prediction is a challenging problem,and accurate prediction of DDI events is critical to patient health and new drug development.Recently,many machine learning-based techniques have been proposed for predicting DDI events.However,most of the existing methods do not effectively integrate the multidimensional features of drugs and provide poor mitigation of noise to get effective feature information.To address these limitations,we propose a DDI-Transform neural network framework for DDI event prediction.In DDI-Transform,we design a drug structure information feature extraction module and a drug bind-protein feature extraction module to obtain multidimensional feature information.A stack of DDI-Transform layers in the DDI-Transform network module are then used for adaptive learning,thus adaptively selecting the effective feature information for prediction.The results show that DDI-Transform can accurately predict DDI events and outperform the state-of-the-art models.Results on different scale datasets confirm the robustness of the method.
文摘Most entity ranking research aims to retrieve a ranked list of entities from a Web corpus given a user query. The rank order of entities is determined by the relevance between the query and contexts of entities. However, entities can be ranked directly based on their relative importance in a document collection, independent of any queries. In this paper, we introduce an entity ranking algorithm named NERank＋. Given a document collection, NERank＋ first constructs a graph model called Topical Tripartite Graph, consisting of document, topic and entity nodes. We design separate ranking functions to compute the prior ranks of entities and topics, respectively. A meta-path constrained random walk algorithm is proposed to propagate prior entity and topic ranks based on the graph model. We evaluate NERank＋ over real-life datasets and compare it with baselines. Experimental results illustrate the effectiveness of our approach.
基金the financial support for this study from the National Natural Science Foundation of China(Nos.52276033 and 92052107).
文摘The present study proposed a shaped sweeping jet(SJ)that possesses the merits of both SJ and shaped hole,which demonstrates significantly improved cooling effectiveness and anti-deposition performance.Compared to a classical 777 shaped hole,the shaped SJ exhibits a maximum enhancement of 70%in cooling effectiveness and a maximum reduction of 28%in particle deposition height,respectively.Owing to the periodic oscillation of coolant jet and higher streamwise jet momentum,the shaped SJ can provide much wider coolant coverage and therefore sweep the adhesive particle away from the wall.This study is the first attempt to reconcile the performance of film cooling and particle anti-deposition simultaneously,which offers a promising design concept for future engine cooling.
文摘In this paper, we proposed a new kind of mark points coded by color and a new quasi-ellipse detector on pixel level. This method is especially applicable to three- dimensional （3D） head panoramic reconstruction. Images of adjacent perspectives can be stitched by matching pasted color-coded mark points in overlap area to calculate the transformation matrix. This paper focuses on how the color-coded mark points work and how to detect and match corresponding points from different perspectives. Tests are performed to show the efficiency and accuracy of this method based on the original data obtained by structured light projection.
