Parallel Clustering Algorithm Research Articles

A scalable multi-density clustering approach to detect city hotspots in a smart city

In the field of Smart City applications, the analysis of urban data to detect city hotspots, i.e., regions where urban events (such as pollution peaks, virus infections, traffic spikes, and crimes) occur at a higher density than in the rest of the dataset, is becoming a common task. The detection of such hotspots can serve as a valuable organizational technique for framing detailed information about a metropolitan area, providing high-level spatial knowledge for planners, scientists, and policymakers. From the algorithmic viewpoint, classic density-based clustering algorithms are very effective in discovering hotspots characterized by homogeneous density; however, their application on multi-density data can produce inaccurate results. For such a reason, since metropolitan cities are characterized by areas with significantly variable densities, multi-density clustering approaches are more effective in discovering city hotspots. Moreover, the growing volumes of data collected in urban environments require the development of parallel approaches, in order to take advantage of scalable executions offered by Edge and Cloud environments. This paper describes the design and implementation of a parallel multi-density clustering algorithm aimed at analyzing high volumes of urban data in an efficient way. The experimental evaluation shows that the proposed parallel clustering approach takes out encouraging advantages in terms of execution time, speedup, and efficiency.

Scalable High-Quality Hypergraph Partitioning

Balanced hypergraph partitioning is an NP-hard problem with many applications, e.g., optimizing communication in distributed data placement problems. The goal is to place all nodes across k different blocks of bounded size, such that hyperedges span as few parts as possible. This problem is well-studied in sequential and distributed settings, but not in shared-memory. We close this gap by devising efficient and scalable shared-memory algorithms for all components employed in the best sequential solvers without compromises with regards to solution quality. This work presents the scalable and high-quality hypergraph partitioning framework Mt-KaHyPar . Its most important components are parallel improvement algorithms based on the FM algorithm and maximum flows, as well as a parallel clustering algorithm for coarsening – which are used in a multilevel scheme with log (n) levels. As additional components, we parallelize the n -level partitioning scheme, devise a deterministic version of our algorithm, and present optimizations for plain graphs. We evaluate our solver on more than 800 graphs and hypergraphs, and compare it with 25 different algorithms from the literature. Our fastest configuration outperforms almost all existing hypergraph partitioners with regards to both solution quality and running time. Our highest-quality configuration achieves the same solution quality as the best sequential partitioner KaHyPar , while being an order of magnitude faster with ten threads. Thus, two of our configurations occupy all fronts of the Pareto curve for hypergraph partitioning. Furthermore, our solvers exhibit good speedups, e.g., 29.6x in the geometric mean on 64 cores (deterministic), 22.3x (log (n) -level), and 25.9x ( n -level).

STRP-DBSCAN: A Parallel DBSCAN Algorithm Based on Spatial-Temporal Random Partitioning for Clustering Trajectory Data

Trajectory clustering algorithms analyze the movement trajectory of the target objects to mine the potential movement trend, regularity, and behavioral patterns of the object. Therefore, the trajectory clustering algorithm has a wide range of applications in the fields of traffic flow analysis, logistics and transportation management, and crime analysis. Existing algorithms do not make good use of the temporal attributes of trajectory data, resulting in a long clustering time and low clustering accuracy of spatial-temporal trajectory data. Meanwhile, the density-based clustering algorithms represented by DBSCAN are very sensitive to the clustering parameters. The radius value Eps and the minimal points number MinPts within Eps radius, defined by the user, have a significant impact on the clustering results, and tuning these parameters is difficult. In this paper, we present STRP-DBSCAN, a parallel DBSCAN algorithm based on spatial-temporal random partitioning for clustering trajectory data. It adopts spatial-temporal random partitioning to distribute balanced computation among different computing nodes and reduce the communication overhead of the parallel clustering algorithm, thus improving the execution efficiency of the DBSCAN algorithm. We also present the PER-SAC algorithm, which uses deep reinforcement learning to combine the prioritized experience replay (PER) and the soft actor-critic (SAC) algorithm for autotuning the optimal parameters of DBSCAN. The experimental results show that STRP-DBSCAN effectively reduces the clustering time of spatial-temporal trajectory data by up to 96.2% and 31.2% compared to parallel DBSCAN and the state-of-the-art RP-DBSCAN. The PER-SAC algorithm also outperforms the state-of-the-art DBSCAN parameter tuning algorithms and improves the clustering accuracy by up to 8.8%. At the same time, the proposed algorithm obtains a higher stability of clustering accuracy.

Distributed structural clustering on large graph

SummaryGraph clustering is a primitive operation for graph data mining. It plays an important role to reveal community clusters, hubs, and outliers in complex networks. There are several graph clustering algorithms have been proposed based on the well‐studied SCAN algorithm in recent years. However, SCAN and its improved sequential variants are prohibitively slow due to their iterative computations. The parallel variants are focusing on improving the efficiency of graph clustering by utilizing multi‐cores computer architectures on single computing node with complex optimization techniques. Therefore, SCAN and its variants are not suitable for processing very large graphs due to the limitations of memory size and storage volume on a single node. In this article, we proposed a distributed parallel structural clustering algorithm using MapReduce. In order to improve the efficiency further, we proposed optimization techniques including partition based clustering and simplified combination with labels to accelerate the operations. We conducted extensive experiments on real world datasets. The experimental results showed our algorithm is high efficiency and scales well under different settings.

A parallel CF tree clustering algorithm for mixed-type datasets

Many traditional clustering algorithms are incapable of processing mixed-type datasets in parallel, limiting their applications in big data. In this paper, we propose a CF tree clustering algorithm based on MapReduce to handle mixed-type datasets. Mapper phase and reducer phase are the two primary phases of MR-CF. In the mapper phase, the original CF tree algorithm is modified to collect intermediate CF entries, and in the reducer phase, k-prototypes is extended to cluster CF entries. To avoid the high costs associated with I/O overheads and data serialization, MR-CF loads a dataset from HDFS only once. We first analyze the time complexity, space complexity, and I/O complexity of MR-CF. We also compare it with sklearn BIRCH, Apache Mahout k-means, k-prototypes, and mrk-prototypes on several real-world datasets and synthetic datasets. Experiments on two mixed-type big datasets reveal that MR-CF reduces execution time by 45.4% and 61.3% when compared to k-prototypes, and it reduces execution time by 73.8% and 55.0% when compared to mrk-prototypes.

Large Data Oriented to Image Information Fusion Spark and Improved Fruit Fly Optimization Based on the Density Clustering Algorithm

The density-based applied spatial clustering algorithm is an algorithm based on high-density interconnected regions, which discovers class clusters of arbitrary shapes in noisy data sets and is widely used. However, it suffers from slow computation speed due to large-scale disk I/O and clustering bias due to uneven density class clusters and poor parameter search ability. To address these problems, a parallel density clustering algorithm based on an improved fruit fly optimization algorithm and Spark memory iteration is proposed. The proposed algorithm first divides the data grid using an irregular dynamic density region partitioning strategy. Then, a hybrid fruit fly particle swarm algorithm based on a genetic optimization mechanism is proposed to achieve dynamic optimization seeking for parameters in local clustering to improve the clustering effect of local clustering. Finally, the local merging of samples in irregularly bounded grid cells under each partition is achieved by designing a custom clustering merging strategy. The experiments show that the improved algorithm is generally applicable to the clustering of different shape class clusters and larger scale data and has obvious improvement in accuracy and parallel efficiency.

CLUE: a clustering algorithm for current and future experiments

CLUE (CLUstering of Energy) is a fast parallel clustering algorithm for High Granularity Calorimeters in High Energy Physics. In these types of detectors, the standard clustering algorithms using combinatorics are expected to fail due to large number of digitized energy deposits (hits) in the reconstruction stage bringing to a consequent memory/timing explosion. This innovative algorithm uses a grid spatial index for fast querying of neighbors and its timing scales linearly with the number of hits within the range considered. Initially CLUE was developed in a standalone repository that allows performance benchmarking with respect to its CPU and GPU implementations, demonstrating the power of algorithmic parallelization in the coming era of heterogeneous computing. CLUE has been successfully used in simulation and beam tests of the High Granularity Calorimeter to be installed for the upgrade of the CMS detector in Phase-2 of the HL-LHC. Recently CLUE was also imported in the key4hep framework and first results will be shown for detectors proposed in future collider projects.

Parallel progressive-based inductive subspace and fuzzy-based firefly algorithm for high ensemble data clustering

Recently, the framework is proposed with techniques such as random subspace and constraint propagation for handling high dimensional data ensemble clustering. Huge dataset clustering is difficult for conventional sequential clustering methods since it requires higher computation time. Distributed parallel processing and methods are consequently useful towards attaining results and scalability constraints of clustering huge datasets. So, in this work, the parallel progressive-based inductive subspace ensemble clustering (PPISEC) algorithm is introduced with the concept of MapReduce (MR) to perform high dimensional data clustering. Depending on micro-clusters and correspondence relative, the clustering method is designed which is easily parallelised via MR and completed in moderately with a small number of MR rounds. However, in the PPISEC algorithm, the centroid values are selected with the help of an improved support vector machine (ISVM) classifier. Thus, the incremental ensemble member chosen (IEMC) progression is performed with fuzzy-based firefly algorithm (FFA), and the normalised cut algorithm is established to accomplish high dimensional data clustering. The outcome shows that, the proposed PPISEC framework, which performs well on three benchmark samples by utilising high dimensionality and enhanced the results than the conventional clustering ensemble approaches.

Parallel DBSCAN Clustering Algorithm Using Hadoop Map-reduce Framework for Spatial Data

Data clustering is the first step for future applications of big data analysis. It is a driving model for Artificial Intelligence and Machine Learning architectures. Processing large volumes of data in faster mode is a big challenge in these applications. which requires fast and efficient algorithms for handling big data. Parallel clustering algorithms are one promising design, which increases the speed of handling such big data. In this paper, a parallel algorithm for clustering a spatial dataset called the P-DBSCAN algorithm is implemented using Hadoop map-reduce framework. This research paper signifies the improvement for data clustering in data analytic applications. The new P-DBSCAN algorithm is executed over generated dataset. The result of this parallel algorithm is compared with existing DBSCAN algorithm to show improvement of runtime performance. This work offers an increase in the performance of execution time. In addition, the outcome of P-DBSCAN shows how to resolve the scalability problem of a large data set.

A new parallel adaptive clustering and its application to streaming data

This paper presents a parallel adaptive clustering (PAC) algorithm to automatically classify data while simultaneously choosing a suitable number of classes. Clustering is an important tool for data analysis and understanding in a broad set of areas including data reduction, pattern analysis, and classification. However, the requirement to specify the number of clusters in advance and the computational burden associated with clustering large sets of data persist as challenges in clustering. We propose a new parallel adaptive clustering (PAC) algorithm that addresses these challenges by adaptively computing the number of clusters and leveraging the power of parallel computing. The algorithm clusters disjoint subsets of the data on parallel computation threads. We develop regularized set k-means to efficiently cluster the results from the parallel threads. A refinement step further improves the clusters. The PAC algorithm offers the capability to adaptively cluster data sets which change over time by reusing the information from previous time steps to decrease computation. We provide theoretical analysis and numerical experiments to characterize the performance of the method, validate its properties, and demonstrate the computational efficiency of the method.

A Study on Parallel Clustering Algorithms based on MapReduce

Since the advent of the global computerized market, the volume of digital information has grown exponentially, as has the demand for storing it. As the price of storage devices decreases, the necessity to analyze vast quantities of unstructured digital data to retain only essential information increases. MapReduce is a programming paradigm for producing and generating massive information indices. Using MapReduce to produce meaningful clusters from such a massive amount of raw data is an efficient way to manage such voluminous amounts of data. On the other hand, the existing industry standard for data clustering algorithms presents significant obstacles. The conventional clustering calculation efficiently handles a great deal of information from various sources, such as online media, business, and the web. Nevertheless, the sequential count in clustering approaches is time-intensive in these conventional calculations. The wide varieties of K-Means, including K-Harmonic Means, are sensitive to forming cluster centers in huge datasets. This work suggests a logical evaluation of such calculations. It offers a study of the various k-means clustering algorithms employed in MapReduce, as well as the study on the introduction and the open challenges of parallelism in MapReduce.

PLOC++

We propose a novel version of the GPU-oriented massively parallel locally-ordered clustering (PLOC) algorithm for constructing bounding volume hierarchies (BVHs). Our method focuses on removing the weaknesses of the original approach by simplifying and fusing different phases, while replacing most performance critical parts by novel and more efficient algorithms. This combination allows for outperforming the original approach by a factor of 1.9 - 2.3x.

A Parallel Fractional Lion Algorithm for Data Clustering Based on MapReduce Cluster Framework

This work introduces a parallel clustering algorithm by modifying the existing Fractional Lion Algorithm (FLA). The proposed work replaces the conventional Euclidean distance measure with the Bhattacharya distance measure to newly propose the improved FLA (IMR-FLA). The proposed IMR-FLA is implemented in both the mapper and the reducer in the MapReduce framework to achieve the parallel clustering. The experimentation of the proposed IMR-FLA is done by using six standard databases, namely Pima Indian diabetes dataset, Heart disease dataset, Hepatitis dataset, localization dataset, breast cancer dataset, and skin segmentation dataset, from the UCI repository. The proposed IMR-FLA has the overall improved Jaccard coefficient value of 0.9357, 0.6572, 0.7462, 0.5944, 0.9418, and 0.8680, for each dataset. Similarly, the proposed IMR-FLA algorithm has outclassed other classifiers' performance with the clustering accuracy value of 0.9674, 0.9471, 0.9677, 0.777, 0.9023, and 0.9585, respectively, for the experimental databases.

Performance Evaluation of Simple K-Mean and Parallel K-Mean Clustering Algorithms: Big Data Business Process Management Concept

Data is the most valuable asset in any firm. As time passes, the data expands at a breakneck speed. A major research issue is the extraction of meaningful information from a complex and huge data source. Clustering is one of the data extraction methods. The basic K-Mean and Parallel K-Mean partition clustering algorithms work by picking random starting centroids. The basic and K-Mean parallel clustering methods are investigated in this work using two different datasets with sizes of 10000 and 5000, respectively. The findings of the Simple K-Mean clustering algorithms alter throughout numerous runs or iterations, according to the study, and so iterations differ for each run or execution. In some circumstances, the clustering algorithms’ outcomes are always different, and the algorithms separate and identify unique properties of the K-Mean Simple clustering algorithm from the K-Mean Parallel clustering algorithm. Differentiating these features will improve cluster quality, lapsed time, and iterations. Experiments are designed to show that parallel algorithms considerably improve the Simple K-Mean techniques. The findings of the parallel techniques are also consistent; however, the Simple K-Mean algorithm’s results vary from run to run. Both the 10,000 and 5000 data item datasets are divided into ten subdatasets for ten different client systems. Clusters are generated in two iterations, i.e., the time it takes for all client systems to complete one iteration (mentioned in chapter number 4). In the first execution, Client No. 5 has the longest elapsed time (8 ms), whereas the longest elapsed time in the following iterations is 6 ms, for a total elapsed time of 12 ms for the K-Mean clustering technique. In addition, the Parallel algorithms reduce the number of executions and the time it takes to complete a task.

Parallel complete gradient clustering algorithm and its properties

Clustering is one of the most important tasks in the field known as ‘Exploratory Data Analysis’ (EDA). It explores the dependencies hidden in individual data attributes, dividing them from one set into smaller subsets. In this paper, a Parallel Complete Gradient Clustering Algorithm (PCGCA) is proposed. The Complete Gradient Clustering Algorithm (CGCA) provides a natural interpretation combined with no need for assumptions regarding the number of clusters, making it an appealing choice. Moreover, in CGCA, internal optimization procedures point out the parameters influencing the size of clusters. Algorithms based on kernel density estimation can, therefore, be applied for diverse practical scenarios. Another very useful usage is outlier detection – which is especially important in the currently fast-growing data industry. The described algorithm has been validated in terms of both the speed of calculation and the quality of the obtained solution. The quality of the solution was evaluated with the use of eleven clustering indexes calculated on six data sets. In addition, the obtained result was compared with several classical well-known methods of clustering.

MR-DBIFOA: a parallel Density-based Clustering Algorithm by Using Improve Fruit Fly Optimization

<p>Clustering is an important technique for data analysis and knowledge discovery. In the context of big data, the density-based clustering algorithm faces three challenging problems: unreasonable division of data gridding, poor parameter optimization ability and low efficiency of parallelization. In this study, a density-based clustering algorithm by using improve fruit fly optimization based on MapReduce (MR-DBIFOA) is proposed to tackle these three problems. Firstly, based on KD-Tree, a division strategy (KDG) is proposed to divide the cell of grid adaptively. Secondly, an improve fruit fly optimization algorithm (IFOA) which use the step strategy based on knowledge learn (KLSS) and the clustering criterion function (CFF) is designed. In addition, based on IFOA algorithm, the optimal parameters of local clustering are dynamically selected, which can improve the clustering effect of local clustering. Meanwhile, in order to improve the parallel efficiency, the density-based clustering algorithm using IFOA (MR-QRMEC) are proposed to parallel compute the local clusters of clustering algorithm. Finally, based on QR-Tree and MapReduce, a cluster merging algorithm (MR-QRMEC) is proposed to get the result of clustering algorithm more quickly, which improve the core clusters merging efficiency of density-based clustering algorithm. The experimental results show that the MR-DBIFOA algorithm has better clustering results and performs better parallelization in big data.</p> <p> </p>

ParChain

This paper studies the hierarchical clustering problem, where the goal is to produce a dendrogram that represents clusters at varying scales of a data set. We propose the ParChain framework for designing parallel hierarchical agglomerative clustering (HAC) algorithms, and using the framework we obtain novel parallel algorithms for the complete linkage, average linkage, and Ward's linkage criteria. Compared to most previous parallel HAC algorithms, which require quadratic memory, our new algorithms require only linear memory, and are scalable to large data sets. ParChain is based on our parallelization of the nearest-neighbor chain algorithm, and enables multiple clusters to be merged on every round. We introduce two key optimizations that are critical for efficiency: a range query optimization that reduces the number of distance computations required when finding nearest neighbors of clusters, and a caching optimization that stores a subset of previously computed distances, which are likely to be reused. Experimentally, we show that our highly-optimized implementations using 48 cores with two-way hyper-threading achieve 5.8--110.1x speedup over state-of-the-art parallel HAC algorithms and achieve 13.75--54.23x self-relative speedup. Compared to state-of-the-art algorithms, our algorithms require up to 237.3x less space. Our algorithms are able to scale to data set sizes with tens of millions of points, which existing algorithms are not able to handle.

Performance improvement of extreme multi-label classification using K-way tree construction with parallel clustering algorithm

eXtreme Multi-Label Classification (XMLC) is the particular case of Multi-Label Classification, which deals with an extremely high number of labels. The main goal is to learn extreme classifier which extracts the subset of relevant labels from extremely large label space. In an extreme environment, one of the big issues is to deal with an extreme number of features, labels and instances which affects the performance of the classifier. The high dimensional feature space and label space makes existing approaches intractable in terms of data scalability, data sparsity, training and prediction cost. The appropriate input representation technique can be used to maintain the interdependency among labels and correlation between feature space and label space. The proposed approach called “K-way Tree based eXtreme Multi-Label Classifier (KTXMLC)“ works on tree based classifier to maintain correlations using feature-label input representation technique and node partitioning using a clustering algorithm. KTXMLC constructs multi-way multiple trees using a parallel clustering algorithm, which leads to fast computational cost. KTXMLC outperforms over the existing tree based classifier in terms of ranking based measures on six datasets named Delicious, Mediamill, Eurlex-4K, Wiki10-31K, AmazonCat-13K, Delicious-200K.

Parallel FCM clustering algorithm of fuzzy number based on cut set

Aiming at the problem of complex method and low efficiency of fuzzy numbers in classification processing, a parallel Fuzzy CMeans (FCM) clustering method based on cut set is proposed. Firstly, according to the decomposition theorem, the fuzzy numbers are divided horizontally into the form of the union of interval numbers, and then the interval numbers are transformed into the determined “real” data, and the parallel FCM clustering algorithm is used to classify the fuzzy numbers. The theoretical analysis and application show that the method has good classification accuracy and efficiency for fuzzy data clustering.

A real-time FPGA-based cluster finding algorithm for LHCb silicon pixel detector

Starting from the next LHC run, the upgraded LHCb High Level Trigger will process events at the full LHC collision rate (averaging 30 MHz). This challenging goal, tackled using a large and heterogeneous computing farm, can be eased addressing lowest-level, more repetitive tasks at the earliest stages of the data acquisition chain. FPGA devices are very well-suited to perform with a high degree of parallelism and efficiency certain computations, that would be significantly demanding if performed on general-purpose architectures. A particularly time-demanding task is the cluster-finding process, due to the 2D pixel geometry of the new LHCb pixel detector. We describe here a custom highly parallel FPGA-based clustering algorithm and its firmware implementation. The algorithm implementation has shown excellent reconstruction quality during qualification tests, while requiring a modest amount of hardware resources. Therefore it can run in the LHCb FPGA readout cards in real time, during data taking at 30 MHz, representing a promising alternative solution to more common CPU-based algorithms.