Search

In this paper we consider Susceptible \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rightarrow $$\end{document}→ Infectious \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\rightarrow $$\end{document}→ Recovered (SIR) epidemics on random graphs with clustering. To incorporate group structure of the underlying social network, we use a generalized version of the configuration model in which each node is a member of a specified number of triangles. SIR epidemics on this type of graph have earlier been investigated under the assumption of homogeneous infectivity and also under the assumption of Poisson transmission and recovery rates. We extend known results from literature by relaxing the assumption of homogeneous infectivity both in individual infectivity and between different kinds of neighbours. An important special case of the epidemic model analysed in this paper is epidemics in continuous time with arbitrary infectious period distribution. We use branching process approximations of the spread of the disease to provide expressions for the basic reproduction number \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R_0$$\end{document}R0, the probability of a major outbreak and the expected final size. In addition, the impact of random vaccination with a perfect vaccine on the final outcome of the epidemic is investigated. We find that, for this particular model, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R_0$$\end{document}R0 equals the perfect vaccine-associated reproduction number. Generalizations to groups larger than three are discussed briefly.

Inferring the magnitude and occurrence of real-world events from natural language text is a crucial task in various domains. Particularly in the domain of public health, the state-of-the-art document and token centric event detection approaches have not kept the pace with the growing need for more robust event detection in public health. In this paper, we propose UPHED, a unified approach, which combines both the document and token centric event detection techniques in an unsupervised manner such that events which are: rare (aperiodic); reoccurring (periodic) can be detected using a generative model for the domain of public health. We evaluate the efficiency of our approach as well as its effectiveness for two real-world case studies with respect to the quality of document clusters. Our results show that we are able to achieve a precision of 60% and a recall of 71% analyzed using manually annotated real-world data. Finally, we also make a comparative analysis of our work with the well-established rule-based system of MedISys and find that UPHED can be used in a cooperative way with MedISys to not only detect similar anomalies, but can also deliver more information about the specific outbreak of reported diseases.

International audience; This paper targets multicast packet-switched transmission on the downlink radio interface using the multicast mode. In the multicast mode, a packet is sent simultaneously to several terminals in the same cell. In order to increase the received SNR, we consider different techniques of macro diversity, namely Selective Combining (SC) and Maximal Ratio Combining (MRC). We develop an analytical model that allows the computation of the throughput for different multicast schedulers with SC and MRC schemes. These schedulers offer different trade-offs between throughput and fairness. We also introduce an efficient user clustering strategy considering their average radio channel quality. The study shows that macro diversity improves the transmission performance especially for pure multicast.

The demand for increasing airport capacity combined with many constraints as well as the complexity of the data itself leads to the use of heuristic methods from the computational intelligence domain. More specifically, the focus in this paper is on how (fuzzy) clustering methods and evolutionary algorithms are applied on various aspects of the Air Traffic Management domain. Fuzzy clustering techniques have been used for data evaluation and pre-processing. One task is the identification and correction of noise and outliers in radar tracks as a pre-processing step. In addition, clustering has been applied to identify general flight routes in retrospective analysis tasks as well as to generate fuzzy rules, thus verifying or complementing expert knowledge regarding transfer passenger movements. Evolutionary algorithms are used to assist air- and ground traffic controllers. Namely in Rogena (free ROuting with GENetic Algorithms) for route planning and TRACC (Taxi Routes for Aircraft: Creation and Controlling) for ground movement planning. Both systems create conflict free routes for aircraft which are suggested to the air- and ground traffic controllers, respectively.

Recent technological advances have increased the quantity of movement data being recorded. While valuable knowledge can be gained by analysing such data, its sheer volume creates challenges. Geovisual analytics, which helps the human cognition process by using tools to reason about data, offers powerful techniques to resolve these challenges. This paper introduces such a geovisual analytics environment for exploring movement trajectories, which provides visualisation interfaces, based on the classic space-time cube. Additionally, a new approach, using the mathematical description of motion within a space-time cube, is used to determine the similarity of trajectories and forms the basis for clustering them. These techniques were used to analyse pedestrian movement. The results reveal interesting and useful spatiotemporal patterns and clusters of pedestrians exhibiting similar behaviour.

"\"This paper proposes a Dynamic Clustering Algorithm as a new regionalization. method for spatial functional data. The method looks for the best partition. optimizing a criterion of spatial association among functional data. Furthermore it. is such that a summary of the variability structure of each cluster is discovered. The. performance of the proposal is checked through an application on real data.\""

The paper describes a system able to recognize the users identity according how she/he looks at the monitor while using a given interface. The system does not need invasive measurements that could limit the naturalness of her/his actions. The proposed approach clusters the sequences of observed points on the screen and characterizes the user identity according the relevant detected patterns. Moreover, the system is able to identify patterns in order to have a more accurate recognition and to create prototypes of natural facial dynamics in user expressions. The possibility to characterize people through facial movements introduces a new perspective on human-machine interaction. For example, a user can obtain different contents according her/his mood or a software interface can modify itself to keep a higher attention from a bored user. The success rate of the classification using only 7 parameters is around 68%. The approach is based on k-means that is tuned to maximize an index involving the number of true-positive detections and conditional probabilities. A different evaluation of this parameter allows to focus on the identification of a single user or to spot a general movement for a wide range of people The experiments show that the performance can reach the 90% of correct recognition.

The discovery of predictive models for process performances is an emerging topic, which poses a series of difficulties when considering complex and flexible processes, whose behaviour tend to change over time depending on context factors. We try to face such a situation by proposing a predictive-clustering approach, where different context-related execution scenarios are equipped with separate prediction models. Recent methods for the discovery of both Predictive Clustering Trees and state-aware process performance predictors can be reused in the approach, provided that the input log is preliminary converted into a suitable propositional form, based on the identification of an optimal subset of features for log traces. In order to make the approach more robust and parameter free, we also introduce an ensemble-based clustering method, where multiple PCTs are learnt (using different, randomly selected, subsets of features), and integrated into an overall model. Several tests on real-life logs confirmed the validity of the approach. © 2013 Springer-Verlag.

International audience; In this paper, we introduce a new clustering strategy for temporally ordered data streams, which is able to discover groups of homogeneous streams performing a single pass on data. It is a two steps approach where an on-line algorithm computes statistics about the dissimilarities among data and then, an off-line algorithm computes the final partition of the streams. The effectiveness of the proposal is evaluated through tests on real data.

"In recent years the analysis of data streams has received a lot of attention.. This is motivated by the increase of the number of applications which generate huge. amounts of high speed temporal data. Let us think to sensor networks, computer networks,. manufactures. Data streams are usually highly evolving, thus mining changes. in data is a challenging task. In this paper we will deal with the structural drift detection. problem where the aim is to discover and to describe changes in proximity. relations among multiple data streams. We will introduce a new strategy whose effectiveness. is shown through an application on simulated data."

Clustering is a crucial network design approach to enable large-scale wireless sensor networks (WSNs) deployments. A large variety of clustering approaches has been presented focusing on various aspect such as minimizing communication overhead, controlling the network topology etc. Simulations on such protocols are performed using theoretical models that are based on unrealistic assumptions like ideal wireless communication channels and perfect energy consumption estimations. With these assumptions taken for granted, theoretical models claim various performance milestones that cannot be achieved in realistic conditions. In this paper, we design a new clustering protocol that adapts to the changes in the environment and the needs and goals of the user applications. We provide a protocol that is deployable protocol in real WSNs. We apply our protocol in multiple indoors wireless sensor testbeds with multiple experimental scenarios to showcase scalability and trade-offs between network properties and configurable protocol parameters. By analysis of the real world experimental output, we present results that depict a more realistic view of the clustering problem, regarding adapting to environmental conditions and the quality of topology control. Our study clearly demonstrates the applicability of our approach and the benefits it offers to both research & development communities.

In this paper, soft computing and artificial intelligence techniques have been used to define a model for simulating users' decisional process in a transportation system. Through this framework, the variables involved are expressed by approximate or linguistic values, like in the humans' reasoning way, in order to forecast users' mode choice behavior. The model has been specified and calibrated using a set of real life data. Results appear good in comparison with those obtained by a classical random utility based model calibrated with the same data, and the methodology seems promising also in case of different applications in the field of choice behavior simulation.

Part 9: Clustering; International audience; This paper proposes a hybrid method for fast and accurate Nearest Neighbor Classification. The method consists of a non-parametric cluster-based algorithm that produces a two-level speed-up data structure and a hybrid algorithm that accesses this structure to perform the classification. The proposed method was evaluated using eight real-life datasets and compared to four known speed-up methods. Experimental results show that the proposed method is fast and accurate, and, in addition, has low pre-processing computational cost.

Many organisations need to extract useful information from huge amounts of movement data. One example is found in maritime transportation, where the automated identification of a diverse range of traffic routes is a key management issue for improving the maintenance of ports and ocean routes, and accelerating ship traffic. This paper addresses, in a first stage, the research challenge of developing an approach for the automated identification of traffic routes based on clustering motion vectors rather than reconstructed trajectories. The immediate benefit of the proposed approach is to avoid the reconstruction of trajectories in terms of their geometric shape of the path, their position in space, their life span, and changes of speed, direction and other attributes over time. For clustering the moving objects, an adapted version of the Shared Nearest Neighbour algorithm is used. The motion vectors, with a position and a direction, are analysed in order to identify clusters of vectors that are moving towards the same direction. These clusters represent traffic routes and the preliminary results have shown to be promising for the automated identification of traffic routes with different shapes and densities, as well as for handling noise data., MOVE EU Cost Action IC0903 (Knowledge Discovery from Moving Objects)

In this paper, we among functional data. A first strategy aims to classify curves spatially dependent and to obtain a spatio-functional model prototype for each cluster. It is based on a Dynamic Clustering Algorithm with on an optimization problem that minimizes the spatial variability among the curves in each cluster. A second one looks simultaneously for an optimal partition of spatial functional data set and a set of bivariate functional regression models associated to each cluster. These models take into account both the interactions among different functional variables and the spatial relations among the observations.

Holter electrocardiographic (ECG) signals are ambulatory long-term registers used to detect heart diseases. These signals normally include more, than one channels and its duration is up to 24 hours. The principal problem of the cardiologists is the manual inspection of the whole Holter ECG to find all those beats which morphology differs from the normal beats. In this paper we present our method. We apply firstly a grid clustering technique. Secondly we use a special density-based clustering algorithm, named Optics. Then we visualize every heart beat in the record, heartbeats in a cluster, furthermore we represent every cluster with median of heartbeats. We can perform manual and further automatic clustering. With this method, the result is easy ECG analysis and optimization of time of processing.

"\"In this paper we propose an extended version of a model-based strategy for clustering spatial functional data. The strategy, we refer, aims simultaneously to classify spatially dependent curves and to obtain a spatial functional model prototype for each cluster. The fit of these models implies to estimate a variogram function, the trace variogram function. Our proposal is to introduce an alternative estimator for the trace-variogram function: a kernel variogram estimator. This works better to adapt spatial varying features of the functional data pattern. Experimental comparisons show this approach has some advantages over the previous one. \""

Clustering is one of the most well known activities in scientific investigation and the object of research in many disciplines, ranging from Statistics to Computer Science. Following Handl et al., it can be summarized as a three step process: (a) choice of a distance function; (b) choice of a clustering algorithm; (c) choice of a validation method. Although such a purist approach to clustering is hardly seen in many areas of science, genomic data require that level of attention, if inferences made from cluster analysis have to be of some relevance to biomedical research. Unfortunately, the high dimensionality of the data and their noisy nature makes cluster analysis of genomic data particularly difficult. This paper highlights new findings that seem to address a few relevant problems in each of the three mentioned steps, both in regard to the intrinsic predictive power of methods and algorithms and their time performance. Inclusion of this latter aspect into the evaluation process is quite novel, since it is hardly considered in genomic data analysis.

The need for integration of different data in the understanding and characterization of reservoirs is continuously growing in petroleum geology. The large amount of data for each well and the presence of different wells to be simultaneously analyzed make this task both complex and time consuming. In this scenario, the development of reliable interpretation methods is of prime importance in order to help the geologist and reduce the subjectivity of data interpretation. In this paper, we propose a novel interpretation method based on the integration of unsupervised and supervised learning techniques. This method uses an unsupervised learning algorithm to objectively and quickly evaluate a large dataset made of subsurface data from different wells in the same field. Then it uses a supervised learning algorithm to predict and propagate the characterization over new wells. To test our approach, we use first hierarchical clustering to then feed several supervised learning algorithms in the classification domain (e.g. decision trees and linear regression).

"In recent years, data streams analysis has gained a lot of attention due to the growth of applicative fields generating huge amount of temporal data. In this paper we will focus on the clustering of multiple streams. We propose a new strategy which aims at grouping similar streams and, together, at computing summaries of the incoming data. This is performed by means of a divide and conquer approach where a continuously updated graph collects information on incoming data and an off-line partitioning algorithm provides the final clustering structure. An application on real data sets corroborates the effectiveness of the proposal. "

In recent years we have seen a vast increase in the volume of information published on weblog sites and also the creation of new web technologies where people discuss actual events. The need for automatic tools to organize this massive amount of information is clear, but the particular characteristics of weblogs such as shortness and overlapping vocabulary make this task difficult. In this work, we present a novel methodology to cluster weblog posts according to the topics discussed therein. This methodology is based on a generative probabilistic model in conjunction with a Self-Term Expansion methodology. We present our results which demonstrate a considerable improvement over the baseline.

This paper deals with the clustering of complex data. The input elements to be clustered are linear models estimated on samples arising from several sub-populations (typologies of individuals). We review the main approaches to the computation of metrics between linear models. We propose to use a Wasserstein based metric for the first time in this field. We show the properties of the proposed metric and an application to real data using a dynamic clustering algorithm.

Practical applications in marketing research often involve mixtures of categorical and continuous variables. For the purpose of clustering, a variety of algorithms has been proposed to deal with mixed mode data. In this paper we apply some of these techniques on two data sets regarding marketing problems. We also propose an approach based on the consensus between partitions obtained by considering separately each variable or subsets of variables having the same scale. This approach may be applied to data with many categorical variables and does not impose restrictive assumptions on the variable distribution. We finally suggest a summarizing fuzzy partition with membership degrees obtained as a function of the classes determined by the different methods.

In recent years curve clustering problem has been handled in several applicative fields. However, most of the proposed approaches are sensitive to outliers. This paper aims to deal with this problem in order to make a partition, obtained by using a Dynamic Curve Clustering Algorithm with free knots spline estimation, more robust. The approach is based on a leave-some-out strategy, which defines a rule on the distances distribution of the curves from the barycenters, in order to identify outliers regions. The method is validated by an application on real data.

In this paper, we classify, review, and experimentally compare major methods that are exploited in the definition, adoption, and utilization of element similarity measures in the context of XML schema matching. We aim at presenting a unified view which is useful when developing a new element similarity measure, when implementing an XML schema matching component, when using an XML schema matching system, and when comparing XML schema matching systems.

Generally speaking, an optimization problem consists in maximization or minimization of some function (objective function) f : S → R. The feasible set S ⊆ Rn can be either finite or infinite, and can be described with the help of a finite or infinite number of equalities and inequalities or in the form of some topological structure in Rn. The methods for solution of a certain optimization problem depend mainly on the properties of the objective function and the feasible set. In this paper, we discuss how specific optimization methods of optimization can be used in some specific areas of data mining, namely, in classification and clustering that are considered interrelated [11]

In this paper the use of topological clustering for decoding population neuronal responses and reducing stimulus features is described. The discrete spike trains, recorded in rat somatosensory cortex in response to sinusoidal vibrissal stimulations characterised by different frequencies and amplitudes, are first interpreted to continuous temporal activities by convolving with a decaying exponential filter. Then the self-organising map is utilised to cluster the continuous responses. The result is a topologically ordered clustering of the responses with respect to the stimuli. The clustering is formed mainly along the product of amplitude and frequency of the stimuli. Such grouping agrees with the energy coding result obtained previously based on spike counts and mutual information. To further investigate how the clustering preserves information, the mutual information between resulting stimulus grouping and responses has been calculated. The cumulative mutual information of the clustering resembles closely that of the energy grouping. It suggests that topological clustering can naturally find underlying stimulus-response patterns and preserve information among the clusters.

Symbolic Data Analysis (SDA) aims to to describe and analyze complex and structured data extracted, for example, from large databases. Such data, which can be expressed as concepts, are modeled by symbolic objects described by multivalued variables. In the present paper we present a new distance, based on the Wasserstein metric, in order to cluster a set of data described by distributions with finite continue support, or, as called in SDA, by “histograms”. The proposed distance permits us to define a measure of inertia of data with respect to a barycenter that satisfies the Huygens theorem of decomposition of inertia. We propose to use this measure for an agglomerative hierarchical clustering of histogram data based on the Ward criterion. An application to real data validates the procedure.

Evaluation of clustering partitions is a crucial step in data processing. A multitude of measures exists, which – unfortunately – give for one data set various results. In this paper we present a visualization technique to visualize single clusters of high-dimensional data. Our method maps single clusters to the plane trying to preserve membership degrees that describe a data point's gradual membership to a certain cluster. The resulting scatter plot illustrates separation of the respecting cluster and the need of additional prototypes as well. Since clusters will be visualized individually, additional prototypes can be added locally where they are needed.

In this paper a crossed clustering algorithm is proposed to partitioning a set of symbolic objects in a fixed number of classes. This algorithm allows, at the same time, to determine a structure (taxonomy) on the categories of the object descriptors. This procedure is an extension of the classical simultaneous clustering algorithms, proposed on binary and contingency tables. It is based on a dynamical clustering algorithm on symbolic objects. The optimized criterion is the Φ2 distance computed between the objects description, given by modal variables (distributions) and the prototypes of the classes, described by marginal profiles of the objects set partitions. The convergence of the algorithm is guaranteed at a stationary value of the criterion, in correspondence of the best partition of the symbolic objects in r classes and the best partition of the symbolic descriptors in c groups. An application on web log data has allowed to validate the procedure and suggest it as an useful tool in the Web Usage Mining context. Fulltext Preview

Clustering of sequential or temporal data is more challenging than traditional clustering as dynamic observations should be processed rather than static measures. This paper proposes a HiddenMarkov Model (HMM)-based technique suitable for clustering of data sequences. The main aspect of the work is the use of a probabilistic model-based approach using HMM to derive new proximity distances, in the likelihood sense, between sequences. Moreover, a novel partitional clustering algorithm is designed which alleviates computational burden characterizing traditional hierarchical agglomerative approaches. Experimental results show that this approach provides an accurate clustering partition and the devised distance measures achieve good performance rates. The method is demonstrated on real world data sequences, i.e. the EEG signals due to their temporal complexity and the growing interest in the emerging field of Brain Computer Interfaces.

International audience; Clustering multivariate data that are contaminated by noise is a complex issue, particularly in the framework of mixture model estimation because noisy data can significantly affect the parameters estimates. This paper addresses this problem with respect to likelihood maximization using the Expectation-Maximization algorithm. Two different approaches are compared. The first one consists in defining mixture models that take into account noise. The second one is based of robust estimation of the model parameters in the maximization step of EM. Both have been tested separately, then jointly. Finally, a hybrid model is proposed. Results on artificial data are given and discussed.