Littérature scientifique sur le sujet « Intracluster Similarity »

Background: Recent advances in the field of information and social network has led to the problem of community detection that has got much attention among the researchers. Objective: This paper focus on community discovery, a fundamental task in network analysis by balancing both attribute and structural similarity. The attribute similarity is evaluated using the Jaccard coefficient and Structural similarity is achieved through modularity. Methods: The proposed algorithm is designed for identifying communities in social networks by fusing attribute and structural similarity. The algorithm retains the node which has high influence on the other nodes within the neighbourhood and subsequently groups the objects based on the similarity of the information among the nodes. The extensive analysis is performed on real world datasets like Facebook, DBLP, Twitter and Flickr with different sizes that demonstrates the effectiveness and efficiency of the proposed algorithm over the other algorithms. Results: The results depicts that the generated clusters have a good balance between the structural and attribute with high intracluster similarity and less intracluster similarity. The algorithm helps to achieve faster runtime for moderately-sized datasets and better runtime for large datasets with superior clustering quality.

Abstract In a galaxy cluster, the relative spatial distributions of dark matter, member galaxies, gas, and intracluster light (ICL) may connote their mutual interactions over the cluster’s evolution. However, it is a challenging problem to provide a quantitative measure for matching the shapes between two multidimensional scalar distributions. We present a novel methodology, named the weighted overlap coefficient (WOC), to quantify the similarity of two-dimensional spatial distributions. We compare the WOC with a standard method known as the modified Hausdorff distance (MHD) method. We find that our method is robust, and performs well even with the existence of multiple substructures. We apply our methodology to search for a visible component whose spatial distribution resembles that of dark matter. If such a component could be found to trace the dark-matter distribution with high fidelity for more relaxed galaxy clusters, then the similarity of the distributions could also be used as a dynamical stage estimator of the cluster. We apply the method to six galaxy clusters at different dynamical stages, simulated within a GRT simulation, which is an N-body simulation using the galaxy replacement technique. Among the various components (stellar particles, galaxies, ICL), the ICL+brightest cluster galaxy (BCG) component most faithfully traced the dark-matter distribution. Among the sample galaxy clusters, the relaxed clusters show stronger similarity in the spatial distribution of the dark matter and ICL+BCG than the dynamically young clusters, while the results of the MHD method show a weaker trend with the dynamical stages.

Reference to geographic scale and geographic space representation are characteristics of geospatial data. This work has discussed two issues related to satellite image data, namely huge size and mixed pixels. In clustering, an unsupervised classification and a set of similar objects are grouped together based on the similarity measures. The similarity between intracluster objects is high, whereas the similarity between intercluster objects is low. This paper proposes a clustering technique called spatial rough k-means that classifies the mixed pixels based on their spatial neighbourhood relationship. The authors compared the performance of different state-of-the-art clustering algorithms with that of proposed algorithms for image partitioning and map-reduce methods. The results show that the proposed algorithm has produced clusters of better quality than state-of-the-art algorithms in both the approaches used for handling the vast input data size. Experiments conducted on Landsat-TM 5 data of Delhi region demonstrate the effectiveness of the proposed work.

The genetic relationships among 10 Australian accessions of Phytophthora vignae (Pv) isolated from cowpea, two Japanese accessions of Pv isolated from adzuki bean and one accession of Phytophthora megasperma f. sp. glycinea (Pmg) isolated from soybean were assessed using Restriction Fragment Length Polymorphisms (RFLPs). Using five high-copy probes derived from genomic libraries of Pv and Pmg, RFLP banding patterns used to calculate the genetic distance, d, between isolates, showed that Pmg was very dissimilar to Pv (d = 0.102). In contrast, a close similarity among different isolates of Pv was observed (maximum d = 0.0039), with most polymorphisms being detected using probe pPmgS63. A dendrogram showed that the Japanese and Australian isolates of Pv each belonged to different clusters, with a similar range of intracluster genetic variation between members of the Australian cluster (d range 0.0001-0.0019) compared to the Japanese cluster (d = 0.0022). Because the genetic distance separating the Australian and Japanese clusters (average d = 0.0033) was less than twice that of the intracluster distances, it is proposed that these two geographically isolated groups of P. vignae have recently been derived from a single (unknown) source, rather than each being indigenous pathogens of their respective hosts cowpea and adzuki bean.

The methodology for analyzing the structure of a web resource using an object model, which is based on the description of the page in HTML and using style sheets, has been proposed. The object of research is a web resource page, the model of which is depicted as a DOM tree. Data on the structural elements of the tree are supplemented with information about the styles of the design of the pages. To determine the similarity of pages, it is proposed to apply a criterion that takes into account the structural and stylistic similarity of pages with the corresponding coefficients. To compare page models with each other, the method of aligning trees will be used. Editing distance is used as a metric, and renaming operations, deleting, and adding a tree node is used as editing operations. To determine the similarity in styles, the Jaccard metric is used. To cluster web pages, the k-means method with a cosine distance measure is applied. Intracluster analysis is carried out using a modification of the Zhang-Shasha algorithm. The proposed approach is implemented in the form of an algorithm and software using Python programming language and related libraries. The computational experiment was performed to analyze the structure of individual websites existing on the Internet, as well as to group pages from different web resources. The structure of the formed clusters was analyzed, the RMS similarity of elements in the middle of the clusters was calculated. To assess the quality of the developed approach for the tasks under consideration, expert partitioning was built, the values of accuracy and completeness metrics were calculated. The results of the analysis of the structure of the web resource can be used to improve the structure of the components of the web resource, to understand the navigation of users on the site, to reengineer the web resource

ABSTRACT X-ray and microwave cluster scaling relations are immensely valuable for cosmological analysis. However, their power is limited by astrophysical systematics that bias mass estimates and introduce additional scatter. Turbulence injected into the intracluster medium via mass assembly contributes substantially to cluster non-thermal pressure support, a significant source of such uncertainties. We use an analytical model to compute the assembly-driven non-thermal pressure profiles of haloes based on Monte Carlo-generated accretion histories. We introduce a fitting function for the average non-thermal pressure fraction profile, which exhibits minimal dependence on redshift at fixed peak height. Using the model, we predict deviations from self-similarity and the intrinsic scatter in the Sunyaev–Zel’dovich effect observable-mass scaling relation (YSZ − M) due solely to inter-cluster variation in mass accretion histories. We study the dependence of YSZ − M on aperture radius, cosmology, redshift, and mass limit. The model predicts $5-9{{\ \rm per\ cent}}$ scatter in YSZ − M at z = 0, increasing as the aperture used to compute YSZ increases from R500c to 5R500c. The predicted scatter lies slightly below that of studies based on non-radiative hydro-simulations, illustrating that assembly history variance is likely responsible for a substantial fraction of scatter in YSZ − M. This should be regarded as a lower bound, which will likely increase with the use of an updated gas density model that incorporates a more realistic response to halo assembly. As redshift increases, YSZ − M deviates more from self-similarity and scatter increases. We show that the YSZ − M residuals correlate strongly with the recent halo mass accretion rate, potentially providing an opportunity to infer the latter.

In the data mining of road networks, trajectory clustering of moving objects plays an important role in many applications. Most existing algorithms for this problem are based on every position point in a trajectory and face a significant challenge in dealing with complex and length-varying trajectories. This paper proposes a grid-based whole trajectory clustering model (GBWTC) in road networks, which regards the trajectory as a whole. In this model, we first propose a trajectory mapping algorithm based on grid estimation, which transforms the trajectories in road network space into grid sequences in grid space and forms grid trajectories by recognizing and eliminating redundant, abnormal, and stranded information of grid sequences. We then design an algorithm to extract initial clustering centers based on density weight and improve a shape similarity measuring algorithm to measure the distance between two grid trajectories. Finally, we dynamically allocate every grid trajectory to the best clusters by the nearest neighbor principle and an outlier function. For the evaluation of clustering performance, we establish a clustering criterion based on the classical Silhouette Coefficient to maximize intercluster separation and intracluster homogeneity. The clustering accuracy and performance superiority of the proposed algorithm are illustrated on a real-world dataset in comparison with existing algorithms.

Since air traffic complexity determines the workload of controllers, it is a popular topic in the research field. Benefiting from deep learning, this paper proposes an air traffic complexity assessment method based on the deep metric of air traffic images. An Ordered Deep Metric (ODM) is proposed to measure the similarity of the ordered samples. For each sample, its interclass loss is calculated to keep it close to the mean of the same class and far from the difference. Then, consecutive samples of the same class are considered as a cluster, and the intracluster loss is calculated to make the samples close to the samples within the same cluster and far from the difference. Finally, we present the ODM-based air traffic complexity assessment method (ATCA-ODM), which uses the ODM results as the input of the classification algorithm to improve the assessment accuracy. We verify our ODM algorithm and ATCA-ODM method on the real traffic dataset of south-central airspace of China. The experimental results demonstrate that the assessment accuracy of the proposed ATCA-ODM method is significantly higher than that of the existing similar methods, which also proves that the proposed ODM algorithm can effectively extract high-dimensional features of the air traffic images.

According to the standard cold dark matter (CDM) cosmology, the structure of dark halos including those of galaxy clusters reflects their mass accretion history. Older clusters tend to be more concentrated than younger clusters. Their structure, represented by the characteristic radius r s and mass M s of the Navarro–Frenk–White (NFW) density profile, is related to their formation time. In this study, we showed that r s , M s , and the X-ray temperature of the intracluster medium (ICM), T X , form a thin plane in the space of ( log r s , log M s , log T X ) . This tight correlation indicates that the ICM temperature is also determined by the formation time of individual clusters. Numerical simulations showed that clusters move along the fundamental plane as they evolve. The plane and the cluster evolution within the plane could be explained by a similarity solution of structure formation of the universe. The angle of the plane shows that clusters have not achieved “virial equilibrium” in the sense that mass/size growth and pressure at the boundaries cannot be ignored. The distribution of clusters on the plane was related to the intrinsic scatter in the halo concentration–mass relation, which originated from the variety of cluster ages. The well-known mass–temperature relation of clusters ( M Δ ∝ T X 3 / 2 ) can be explained by the fundamental plane and the mass dependence of the halo concentration without the assumption of virial equilibrium. The fundamental plane could also be used for calibration of cluster masses.

In movement science with inertial sensor many different methodologies resolving specific aspects of movement recognition have been proposed. They are very interesting, and useful, but none of them are generally explicative of what is going on in the semantic sense. When we go down to the movement recognition/classification area (for example in Ambient Intelligence) we do not have a feasible model that can be considered generally predictive or usable for activity recognition. Also, in the field of movement recognition with inertial sensors many technological issues arise: technological diversity, calibration matters, sensor model problems, orientation and position of sensors, and a lot of numerous specificities that, with all the above aspects, and the lack of public dataset of movements sufficiently generic and semantically rich, contribute to create a strong barrier to any approach to a classification matters with wearable sensors. We have also to notice that a movement is a phenomenon explicitly or implicitly (voluntary or involuntary) controlled by brain. The individual free-will introduce a further matter when we want to temporary predict the movements looking at the close past. Pattern can change at any time when ambient, psychological context, age of the subject change. Also, pathological issues, and physiological differences and the will of the subject, introduce important differences. For all these reasons I considered that a semantical /lexical approach to movement recognition with sensors, driven by machine learning techniques could be a promising way to solve some of these challenge and problems. In this Ph.D. Thesis wearable inertial sensors has been used to classify movements, the choice of inertial sensors has been driven by technological and practical advantages, they are cheap, lightweight, and - differently from video cameras - are not prone to the hidden face, or luminance problems. The main idea is to use inertial sensor to understand what a person is doing for ambient-intelligent, healthcare, medical-sport applications. My principal concerns was to propose a method that was not centered on technology issues but on data analysis, that could be a general framework and could also create a general representation of movement,that could be useful also in other area of research, like reasoning. Inertial sensors are treated just as an example, a particular type of sensors, the method is new, reusable, algorithmically simple, net and easy to understand. Accuracy is very high outperforming the best results given in literature, reducing the error rate of 4 times.

The compatibility issues among the characteristics of data involving numerical as well as categorical attributes (mixed) laid many challenges in pattern recognition field. Clustering is often used to group identical elements and to find structures out of data. However, clustering categorical data poses some notable challenges. Particularly clustering diversified (mixed) data constitute bigger challenges because of its range of attributes. Computations on such data are merely too complex to match the scales of numerical and categorical values due to its ranges and conversions. This chapter is intended to cover literature clustering algorithms in the context of mixed attribute unlabelled data. Further, this chapter will cover the types and state of the art methodologies that help in separating data by satisfying inter and intracluster similarity. This chapter further identifies challenges and Future research directions of state-of-the-art clustering algorithms with notable research gaps.