It is well-known that the values of symbolic variables may take various forms such as an interval, a set of stochastic measurements of some underlying patterns or qualitative multi-values and so on. However, the major...It is well-known that the values of symbolic variables may take various forms such as an interval, a set of stochastic measurements of some underlying patterns or qualitative multi-values and so on. However, the majority of existing work in symbolic data analysis still focuses on interval values. Although some pioneering work in stochastic pattern based symbolic data and mixture of symbolic variables has been explored, it still lacks flexibility and computation efficiency to make full use of the distinctive individual symbolic variables. Therefore, we bring forward a novel hierarchical clustering method with weighted general Jaccard distance and effective global pruning strategy for complex symbolic data and apply it to emitter identification. Extensive experiments indicate that our method has outperformed its peers in both computational efficiency and emitter identification accuracy.展开更多
This paper examines the visualization of symbolic data and considers the challenges rising from its complex structure.Symbolic data is usually aggregated from large data sets and used to hide entry specific details an...This paper examines the visualization of symbolic data and considers the challenges rising from its complex structure.Symbolic data is usually aggregated from large data sets and used to hide entry specific details and to transform huge amounts of data(like big data)into analyzable quantities.It is also used to offer an overview in places where general trends are more important than individual details.Symbolic data comes in many forms like intervals,histograms,categories and modal multi-valued objects.Symbolic data can also be considered as a distribution.Currently,the de facto visualization approach for symbolic data is zoomstars which has many limitations.The biggest limitation is that the default distributions(histograms)are not supported in 2D as additional dimension is required.This paper proposes several new improvements for zoomstars which would enable it to visualize histograms in 2D by using a quantile or an equivalent interval approach.In addition,several improvements for categorical and modal variables are proposed for a clearer indication of presented categories.Recommendations for different approaches to zoomstars are offered depending on the data type and the desired goal.Furthermore,an alternative approach that allows visualizing the whole data set in comprehensive table-like graph,called shape encoding,is proposed.These visualizations and their usefulness are verified with three symbolic data sets in exploratory data mining phase to identify trends,similar objects and important features,detecting outliers and discrepancies in the data.展开更多
A large amount of researches and studies have been recently performed by applying statistical and machine learning techniques for vibration-based damage detection. However, the global character inherent to the limited...A large amount of researches and studies have been recently performed by applying statistical and machine learning techniques for vibration-based damage detection. However, the global character inherent to the limited number of modal properties issued from operational modal analysis may be not appropriate for early-damage, which has generally a local character. The present paper aims at detecting this type of damage by using static SHM data and by assuming that early-damage produces dead load redistribution. To achieve this objective a data driven strategy is proposed, consisting of the combination of advanced statistical and machine learning methods such as principal component analysis, symbolic data analysis and cluster analysis. From this analysis it was observed that, under the noise levels measured on site, the proposed strategy is able to automatically detect stiffness reduction in stay cables reaching at least 1%.展开更多
In data analysis tasks, we are often confronted to very high dimensional data. Based on the purpose of a data analysis study, feature selection will find and select the relevant subset of features from the original fe...In data analysis tasks, we are often confronted to very high dimensional data. Based on the purpose of a data analysis study, feature selection will find and select the relevant subset of features from the original features. Many feature selection algorithms have been proposed in classical data analysis, but very few in symbolic data analysis (SDA) which is an extension of the classical data analysis, since it uses rich objects instead to simple matrices. A symbolic object, compared to the data used in classical data analysis can describe not only individuals, but also most of the time a cluster of individuals. In this paper we present an unsupervised feature selection algorithm on probabilistic symbolic objects (PSOs), with the purpose of discrimination. A PSO is a symbolic object that describes a cluster of individuals by modal variables using relative frequency distribution associated with each value. This paper presents new dissimilarity measures between PSOs, which are used as feature selection criteria, and explains how to reduce the complexity of the algorithm by using the discrimination matrix.展开更多
Nowadays,human activity recognition is becoming a more and more significant topic,and there is also a wide range of applications for it in real world scenarios.Sensor data is an important data source in engineering an...Nowadays,human activity recognition is becoming a more and more significant topic,and there is also a wide range of applications for it in real world scenarios.Sensor data is an important data source in engineering and application.At present,some studies have been carried out in the field of human activity recognition based on sensor data in a macroscopic perspective.However,many studies in this perspective face some limitations.One pivotal limitation is uncontrollable data segment length of different kinds of activities.Suitable feature and data form are also influencing factors.This paper carries out the study creatively on a microscopic perspective with an emphasis on the logic and relevance between data segments,attempting to apply the idea of natural language processing and the method of data symbolization to the study of human activity recognition and try to solve the problem above.In this paper,several activity-element definitions and three algorithms are proposed,including the algorithm of dictionary building,the algorithm of corpus building,and activity recognition algorithm improved from a natural language analysis method,TFIDF.Numerous experiments on different aspects of this model are taken.The experiments are carried out on six complex and representative single-level sensor datasets,namely UCI Sports and Daily dataset,Skoda dataset,WISDM Phoneacc dataset,WISDM Watchacc dataset,Healthy Older People dataset and HAPT dataset,which prove that this model can be applied to different datasets and achieve a satisfactory recognition result.展开更多
基金This work was supported by the National Natural Science Foundation of China under Grant Nos. 61771177 and 61701454, the Natural Science Foundation of Jiangsu Province of China under Grant Nos. BK20160147 and BK20160148, and the Academy Project of Finland under Grant No. 310321.
文摘It is well-known that the values of symbolic variables may take various forms such as an interval, a set of stochastic measurements of some underlying patterns or qualitative multi-values and so on. However, the majority of existing work in symbolic data analysis still focuses on interval values. Although some pioneering work in stochastic pattern based symbolic data and mixture of symbolic variables has been explored, it still lacks flexibility and computation efficiency to make full use of the distinctive individual symbolic variables. Therefore, we bring forward a novel hierarchical clustering method with weighted general Jaccard distance and effective global pruning strategy for complex symbolic data and apply it to emitter identification. Extensive experiments indicate that our method has outperformed its peers in both computational efficiency and emitter identification accuracy.
文摘This paper examines the visualization of symbolic data and considers the challenges rising from its complex structure.Symbolic data is usually aggregated from large data sets and used to hide entry specific details and to transform huge amounts of data(like big data)into analyzable quantities.It is also used to offer an overview in places where general trends are more important than individual details.Symbolic data comes in many forms like intervals,histograms,categories and modal multi-valued objects.Symbolic data can also be considered as a distribution.Currently,the de facto visualization approach for symbolic data is zoomstars which has many limitations.The biggest limitation is that the default distributions(histograms)are not supported in 2D as additional dimension is required.This paper proposes several new improvements for zoomstars which would enable it to visualize histograms in 2D by using a quantile or an equivalent interval approach.In addition,several improvements for categorical and modal variables are proposed for a clearer indication of presented categories.Recommendations for different approaches to zoomstars are offered depending on the data type and the desired goal.Furthermore,an alternative approach that allows visualizing the whole data set in comprehensive table-like graph,called shape encoding,is proposed.These visualizations and their usefulness are verified with three symbolic data sets in exploratory data mining phase to identify trends,similar objects and important features,detecting outliers and discrepancies in the data.
文摘A large amount of researches and studies have been recently performed by applying statistical and machine learning techniques for vibration-based damage detection. However, the global character inherent to the limited number of modal properties issued from operational modal analysis may be not appropriate for early-damage, which has generally a local character. The present paper aims at detecting this type of damage by using static SHM data and by assuming that early-damage produces dead load redistribution. To achieve this objective a data driven strategy is proposed, consisting of the combination of advanced statistical and machine learning methods such as principal component analysis, symbolic data analysis and cluster analysis. From this analysis it was observed that, under the noise levels measured on site, the proposed strategy is able to automatically detect stiffness reduction in stay cables reaching at least 1%.
文摘In data analysis tasks, we are often confronted to very high dimensional data. Based on the purpose of a data analysis study, feature selection will find and select the relevant subset of features from the original features. Many feature selection algorithms have been proposed in classical data analysis, but very few in symbolic data analysis (SDA) which is an extension of the classical data analysis, since it uses rich objects instead to simple matrices. A symbolic object, compared to the data used in classical data analysis can describe not only individuals, but also most of the time a cluster of individuals. In this paper we present an unsupervised feature selection algorithm on probabilistic symbolic objects (PSOs), with the purpose of discrimination. A PSO is a symbolic object that describes a cluster of individuals by modal variables using relative frequency distribution associated with each value. This paper presents new dissimilarity measures between PSOs, which are used as feature selection criteria, and explains how to reduce the complexity of the algorithm by using the discrimination matrix.
基金the National Key R&D Program of China(2019YFB1405302)the NSFC(Grant No.61872072)the State Key Laboratory of Computer Software New Technology Open Project Fund(KFKT2018B05).
文摘Nowadays,human activity recognition is becoming a more and more significant topic,and there is also a wide range of applications for it in real world scenarios.Sensor data is an important data source in engineering and application.At present,some studies have been carried out in the field of human activity recognition based on sensor data in a macroscopic perspective.However,many studies in this perspective face some limitations.One pivotal limitation is uncontrollable data segment length of different kinds of activities.Suitable feature and data form are also influencing factors.This paper carries out the study creatively on a microscopic perspective with an emphasis on the logic and relevance between data segments,attempting to apply the idea of natural language processing and the method of data symbolization to the study of human activity recognition and try to solve the problem above.In this paper,several activity-element definitions and three algorithms are proposed,including the algorithm of dictionary building,the algorithm of corpus building,and activity recognition algorithm improved from a natural language analysis method,TFIDF.Numerous experiments on different aspects of this model are taken.The experiments are carried out on six complex and representative single-level sensor datasets,namely UCI Sports and Daily dataset,Skoda dataset,WISDM Phoneacc dataset,WISDM Watchacc dataset,Healthy Older People dataset and HAPT dataset,which prove that this model can be applied to different datasets and achieve a satisfactory recognition result.