Combination Of Clustering Algorithm Research Articles

Classification of Datasets Based on Combination Algorithm of Clustering and Neural Network

How to get the needed information from the data accurately and effectively for analysis is a hot research topic nowadays. Accurate classification of data is the basis for smooth data analysis. In order to classify data effectively, scholars have proposed some classification algorithms, and the most frequently mentioned one is k-means algorithm. However, in previous studies, scholars have directly determined the number of classes to be classified for the data set to be classified. Therefore, in this paper, a combinatorial algorithm is proposed to improve the classification of data with unknown group classes. The clustering algorithm and neural network are also combined to improve classification accuracy. The main elements of the algorithm proposed in this paper are as follows. First, one-third of a set of unknown group class data is selected as the sample data. In order to accurately assess the characteristics of a set of anonymous group class data, it is important first to choose a sample of the data. A sample of one-third of the total data set should be sufficient to provide a reliable representation of the entire population. This sample should be selected at random in order to ensure that the results of the assessment are as accurate as possible. The number of classes is determined by using hierarchical clustering method on the sample data. Then, the non-hierarchical clustering k-means method is used to classify the sample data. Finally, the classification results are trained as the training items of the neural network, and then the model generated after the training is used to classify the overall data. This paper selects three datasets with different kinds, different numbers of variables, and different amounts of data for the experiments and analysis. This paper presents a comprehensive analysis of three distinct datasets. Each dataset has its own unique characteristics, such as its type, the number of variables, and the amount of data contained therein. By leveraging the properties of these datasets, the experiments, and analysis conducted in this paper will provide valuable insights into the data structures and trends contained within. Furthermore, the results from this analysis will serve as a foundation for further research and experimentation. The experimental results show that the combination of clustering algorithm and neural network algorithm will help to improve the accuracy of data classification and identification effectively. This research provides a new way to accurately and effectively perform data classification. Keywords: hierarchical clustering; k-means; neural network; data classification.

Multi-Scale Recursive Identification of Urban Functional Areas Based on Multi-Source Data

The study of urban functional area identification is of great significance for urban function cognition, spatial planning, and economic development. In the identification of urban functional areas, most studies considered only a single data source and a single division scale, the research results have problems such as low update frequency or incomplete information in a single data set, and overfitting or underfitting in a single spatial resolution. Aiming at the above problems, this paper proposes a multi-scale recursive recognition method based on interactive validation for urban functional areas using taxi trajectory data and point of interest (POI) data as the main data sources. First, the dynamic time warping (DTW) algorithm generates a time series similarity matrix, a CA-RFM model combining the clustering algorithm and random forest model is constructed. The model extracts significant feature regions as inputs through a K-medoid clustering algorithm, which are imported into the random forest model for urban functional zone (UFZ) identification. Then, to overcome the shortcomings of a single scale in expressing urban structural characteristics, a recursive model of different levels of urban road networks is established to classify multi-scale functional areas. Finally, cross-validation using the CA-RFM model and POI quantitative identification method obtains the final identification results of urban functional areas. This paper selects Shenzhen as the study area, the results show that the combination of clustering algorithm and random forest model greatly reduces the error of manual selection of training samples. In addition, the study demonstrates the superiority of the proposed method in two aspects, namely, faster delineation and improved accuracy in urban functional area identification.

MSCA: An Unsupervised Anomaly Detection System for Network Security in Backbone Network

Anomaly detection is a crucial topic in network security which refers to automatically mining known and unknown attacks or threats. Many detectors have been proposed in the last decade. Nonetheless, a practical solution, which is able to provide a high True Positive Rate (TPR) with an acceptable False Positive Rate (FPR) without any prior information, is still challenging due to the complexity and variability of anomaly pattern. In this article, we propose a novel unsupervised detection system called MSCA which applies multiple sketches, K-means++ unsupervised clustering, and association rule mining to detect traffic anomalies and analyze anomalous features and correlations. It can blindly identify known and unknown traffic anomalies without any labeled traffic or prior signatures about data distribution. Rich traffic data is first aggregated and compacted to traffic flows by sketches, and further detected by the combination of clustering algorithm and voting strategy. Then association rule mining is finally utilized to find the anomalous frequent item-sets and association rules. Numerical experiments on MAWILAB datasets demonstrate that the proposed detection method outperforms other reference unsupervised detection methods. It achieves an accuracy of 99.86%, 99.97%, 97.08%, and 95.19% in overall four detection types including IP and port of source and destination.

Identification of miR-203a, mir-10a, and miR-194 as predictors for risk of lymphovascular invasion in head and neck cancers.

Lymphovascular invasion (LVI) is an important prognostic indicator of lymph node metastasis and disease aggressiveness but clear molecular mechanisms mediating this in head and neck cancers (HNSC) remain undefined. To identify important microRNAs (miRNAs) in HNSC that associate with and are also predictive of increased risk of LVI, we used a combination of clustering algorithms, multiple regression analyses and machine learning approaches and analyzed miRNA expression profiles in the TCGA HNSC database. As the first step, we identified miRNAs with increased association with LVI as a binary variable. In order to determine whether the identified miRNAs would show functional clusters that are also indicative of increased risk for LVI, we carried out unsupervised as well as supervised clustering. Our results identified distinct clusters of miRNAs that are predictive of increased LVI. We further refined these findings using a Random forest approach, and miR-203a-3p, mir-10a-5p, and miR-194-5p to be most strongly associated with LVI. Pathway enrichment analysis showed these miRNAs targeted genes involved in Hippo signaling and fatty acid oxidation pathways that are mediators of lymph node metastasis. Specific association was also identified between the miRNAs associated with LVI and expression of several lymphangiogenic genes that could be critical for determination of therapeutic strategies.

A Partition Based Framework for Large Scale Ontology Matching

:Large amount of data coming from different sources and formats is available on the web which generates heterogeneity problem. Semantic web technologies play an important role for collecting, merging, matching and aggregating big data from heterogeneous resources by determining the semantic correspondence between the entities. However, achieving good efficiency is major challenge for large scale ontology matching task.Objective:We propose a PBOM framework for coping with the large scale ontology matching problem.Methods:Our proposal first selects the source ontology and calculates the similarity of concepts within source ontology by using Lin measure. We use clustering algorithm for partition of the source ontology, obtained clusters of source ontology then used to divide the target ontology. During matching process, we run matchers from the pool of the matchers over each clusters. We aggregate the result of element level matchers and structure level matchers according to weighted sum aggregation. Each cluster is executed in its processor in parallel with other clusters.Results:We have presented step wise execution of proposed approach over one cluster of source and target ontology. The evaluation of our framework is performed by OAEI datasets of bibliographic benchmark 2014, biomedical track 2015 and anatomy 2016.Conclusion:Results show that, the performance of our approach is better in term of F-measure. The combination of clustering algorithm and parallel processing reduces the memory space and time complexity of matching process.

A computational approach for detecting micro-domains and confinement domains in cells: a simulation study

In this paper, we aim to detect trapping areas (equivalently microdomains or confinement areas) within cells, corresponding to regions where molecules are trapped and thereby undergo subdiffusion. We propose an original computational approach that takes as input a set of molecule trajectories estimated by appropriate tracking methods. The core of the algorithm is based on a combination of clustering algorithms with trajectory classification procedures able to distinguish subdiffusion, superdiffusion and Brownian motion. The idea is to automatically identify trapping areas where we observe a high concentration of subdiffusive particles. We evaluate our proof of concept on artificial sequences obtained with a biophysics-based simulator (Fluosim), and we illustrate its potential on real TIRF microscopy data.

Creating functional groups of marine fish from categorical traits.

BackgroundFunctional groups serve two important functions in ecology: they allow for simplification of ecosystem models and can aid in understanding diversity. Despite their important applications, there has not been a universally accepted method of how to define them. A common approach is to cluster species on a set of traits, validated through visual confirmation of resulting groups based primarily on expert opinion. The goal of this research is to determine a suitable procedure for creating and evaluating functional groups that arise from clustering nominal traits.MethodsTo do so, we produced a species by trait matrix of 22 traits from 116 fish species from Tasman Bay and Golden Bay, New Zealand. Data collected from photographs and published literature were predominantly nominal, and a small number of continuous traits were discretized. Some data were missing, so the benefit of imputing data was assessed using four approaches on data with known missing values. Hierarchical clustering is utilised to search for underlying data structure in the data that may represent functional groups. Within this clustering paradigm there are a number of distance matrices and linkage methods available, several combinations of which we test. The resulting clusters are evaluated using internal metrics developed specifically for nominal clustering. This revealed the choice of number of clusters, distance matrix and linkage method greatly affected the overall within- and between- cluster variability. We visualise the clustering in two dimensions and the stability of clusters is assessed through bootstrapping.ResultsMissing data imputation showed up to 90% accuracy using polytomous imputation, so was used to impute the real missing data. A division of the species information into three functional groups was the most separated, compact and stable result. Increasing the number of clusters increased the inconsistency of group membership, and selection of the appropriate distance matrix and linkage method improved the fit.DiscussionWe show that the commonly used methodologies used for the creation of functional groups are fraught with subjectivity, ultimately causing significant variation in the composition of resulting groups. Depending on the research goal dictates the appropriate strategy for selecting number of groups, distance matrix and clustering algorithm combination.

Exploring multi-objective trade-offs in the design space of a waste heat recovery system

A waste heat recovery system (WHRS) is used to capture waste heat released from an industrial process, and transform the heat into reusable energy. In practice, it can be difficult to identify the optimal form of a WHRS for a particular installation, since this can depend on various design objectives, which are often mutually exclusive. More so when the number of objectives is large. To address this problem, a multiobjective evolutionary algorithm (MOEA) was used to explore and characterise the trade-off surface within the design space of a particular WHRS. A combination of clustering algorithm and parallel coordinates plots was proposed for use in analysing the results. The trade-off surface is first segmented using a clustering algorithm and parallel coordinates plots are then used to both visualise and understand the resulting set of Pareto-optimal designs. As a case study, a simulation of a WHRS commonly found in the food and drinks process industries was developed, comprising of a desuperheater coupled to a hot water reservoir. The system was parameterised, considering typical objectives, and the MOEA used to build a library of alternative Pareto-optimal designs that can be used by installers. The resulting visualisation are used to better understand the sensitivity of the system’s parameters and their trade-offs, providing another source of information for prospective installations.

Combination of clustering algorithms to maximize the lifespan of distributed wireless sensors

Abstract. Increasing the lifespan of a group of distributed wireless sensors is one of the major challenges in research. This is especially important for distributed wireless sensor nodes used in harsh environments since it is not feasible to replace or recharge their batteries. Thus, the popular low-energy adaptive clustering hierarchy (LEACH) algorithm uses the “computation and communication energy model” to increase the lifespan of distributed wireless sensor nodes. As an improved method, we present here that a combination of three clustering algorithms performs better than the LEACH algorithm. The clustering algorithms included in the combination are the k-means+ + , k-means, and gap statistics algorithms. These three algorithms are used selectively in the following manner: the k-means+ + algorithm initializes the center for the k-means algorithm, the k-means algorithm computes the optimal center of the clusters, and the gap statistics algorithm selects the optimal number of clusters in a distributed wireless sensor network. Our simulation shows that the approach of using a combination of clustering algorithms increases the lifespan of the wireless sensor nodes by 15 % compared with the LEACH algorithm. This paper reports the details of the clustering algorithms selected for use in the combination approach and, based on the simulation results, compares the performance of the combination approach with that of the LEACH algorithm.

PROBABILISTIC LONG RANGE FORECAST SYSTEM FOR OIL SPILL TRAJECTORIES IN THE GULF OF BISCAY (SPAIN)

The increasing number of accidental oil spills has motivated the development and implementation of operational oceanography systems (OOS) to help in the decision process during oil spill emergency situations. Currently, most of the national and regional OOS have been setup for short-term (up to 5 days) oil spill forecast. However, recent accidental oil spills such as Prestige in Spain (2002) or Deep Horizon in Gulf of Mexico (2010), have revealed the importance of having larger prediction horizons (up to 30 days) in regional-scale areas. In this work, we have developed a stochastic methodology based on the combination of clustering algorithms and Markov chains of first order to provide medium term (15–30 days) probabilistic oil spill forecasts. The method encompasses the following steps: (1) classification of representative atmospheric patterns using clustering techniques (PCA and k-means [1]); (2) determination the transition probability matrix associated with the Markov chain. The element of the transition matrix (pij) represents the probability of moving from a cluster “i” to a cluster “j” in one time step. In case an accident occurs, the Markov chain provides through the transition probability matrix, the evolution of ocean-atmospheric conditions during the forecasting period; (3) this result is used to force TESEO Lagrangian transport model [2] which allows the characterization of trajectories in probabilistic terms during the forecasting period. The methodology has been applied in the Gulf of Biscay (Spain) to simulate the evolution of oil slick observations and drifter buoys gathered during the Prestige accident. The cumulative probability maps have been compared with these data (oil slicks observations and drifter data), showing that actual trajectories are consistent with the probability of contamination obtained. Results seem promising and we expect to reduce uncertainty by incorporating autoregressive logistic models to help improving the possible evolution of the ocean-atmospheric conditions. A detailed description of the methodology, application and validation will be shown in the presentation and in the final paper.

Prediction of Course Selection by Student using Combination of Data Mining Algorithms in E-Learning

recommender system aims at predicting the best combination of courses selected by students. Here in this paper we present how the combination of clustering algorithm- Simple K-means Algorithm & association rule algorithm- Apriori Association Rule is useful in Course Recommender system. If we use only the Apriori association rule algorithm then we need to preprocess the data obtained from Moodle database. But if we use this combination of clustering & association rule then there is no need to preprocess the data. So we present this new approach & also present the result. To test the result we have used the open source data mining tool Weka.

Effectiveness of Environmental Cluster Analysis in Representing Regional Species Diversity

A major challenge of regional conservation planning is the identification of sets of sites that together represent the overall biodiversity of the relevant region. Environmental cluster analysis (ECA) has been proposed as a potential tool for efficient selection of conservation sites, but the consequences of methodological decisions involved in its application have not been tested so far. We evaluated the performance of ECA with respect to two such decisions: the choice of the clustering algorithm (single linkage, complete linkage, unweighted arithmetic average, unweighted centroid, Ward's minimum variance, and the ALOC algorithm) and the weight given to different groups of environmental variables (rainfall, temperature, and lithology). Specifically we tested how these decisions affect the spatial configuration of clusters of sites defined by the ECA, whether and how they affect the effectiveness of the ECA (i.e., its ability to represent regional species diversity), and whether the effectiveness of alternative methods of hierarchical clustering can be predicted a priori based on the cophenetic correlation. We used an extensive database of the flora of Israel to test these questions. Differences in both the clustering algorithm and the weighting regime had considerable effects on the spatial configuration of the ECA clusters. The single-linkage algorithm produced mostly single-cell clusters plus a single large-sized cluster and was therefore found inappropriate for environmental regionalization. The effectiveness of the ECA was also sensitive to changes in the clustering algorithm and the weighting regime. Yet, most combinations of clustering algorithms and weighting regimes performed significantly better in capturing regional biodiversity than random null models. The main deviation was classifications based on Ward's minimum variance algorithm, which performed less well relative to all other algorithms. The two algorithms that showed the highest effectiveness (unweighted average and unweighted centroid clustering) also exhibited the highest values of the cophenetic correlation, suggesting that this index may serve as a potential indicator for the effectiveness of alternative ECA algorithms.