Data Clustering

The extraction of numeric features to characterize textures on images takes special relevance in certain satellite and aerial images classification processes. The wide range of the methodological approaches used and their applications in the earth observation (i.e., vegetation analysis, urban distribution and growing, landscape analysis, etc.) makes difficult the appropriate selection of the method in each particular case. In this work, several texture analysis techniques were evaluated: statistical features extracted from the grey level co-ocurrence matrix, methods based on energy filters and the edgeness factor, and the most recent based on the wavelets decomposition theory. In addition, different methodological parameters were tested, the most relevant being the neighbourhood size chosen for the determination of the texture features, as well as the distance between pixels and the number of grey levels used. The wavelets based methods offer a wide range of options, specially regarding the type of mother wavelets used, or if the feature extraction is made from images that result from the standard decomposition or from the decomposition and further reconstruction using the inverse transform. Several types of mother wavelets and both decomposition options were tested. One of the best advantages of this methods is that multiresolution analysis allows the combination of features from different levels of resolution. The evaluation was carried out using a mosaic of real images, with the goal of discriminating between different types of crops. The results show that the combination of the original and wavelet features together yields a better performance when the appropriate wavelet is chosen. The wavelets with highger support, such as the Coif-24, generated better results, and most of the texture information of the tested vegetation classes is contained in the lower levels of decomposition.

Due to the dramatic increase of data volumes in different applications, it is becoming infeasible to keep these data in one centralized machine. It is becoming more and more natural to deal with distributed databases and networks. That is why distributed data mining techniques have been introduced. One of the most important data mining problems is data clustering. While many clustering algorithms exist for centralized databases, there is a lack of efficient algorithms for distributed databases. In this paper, an efficient algorithm is proposed for clustering distributed databases. The proposed methodology employs an iterative optimization technique to achieve better clustering objective. The experimental results reported in this paper show the superiority of the proposed technique over a recently proposed algorithm based on a distributed version of the well known K-Means algorithm (Datta et al. 2009) [1].

Rb-Sr and K-Ar ages have been obtained on six biotites, two muscovites and one hornblende from samples of micaschist, gneiss and amphibolite of Lower Paleozoic to Precambrian age at a depth exceeding 2,000 m in basement rocks of the Larderello-Travale geothermal region. Most of the data cluster in the range 2.5-3.7 Ma, revealing the existence of a Pliocene thermal event to which the origin of the field may be attributed. The resulting duration of the Larderello geothermal field is unexpectedly long. In the basement levels of the two wells examined, unstabilized minimum temperatures of 290 ~ and 380 ~ C were measured. All the biotites show almost complete 4~ and 87Sr retention at the measured well temperatures. Petrologic evidence (stilpnomelane stability) and experimental data (activation energies and diffusion coefficients) also favour a closure temperature above 400 ~ C for Rb-Sr and K-Ar in biotites, in agreement with recent direct experimental determinations. For the last 3 Ma mean geothermal gradients of 120~ ~ C/kin have been evaluated in the first 2-3 km, and 600-65 ~ C/km in the underlying 2 km. A rough estimate of total cooling in the last 3 Ma gives a value of 120 ~ C at 2,500 m depth and 50 ~ C at 4,000 m depth in Sasso 22 well. A mean uplift rate of about 0.2 mm/year is calculated independently.

Cluster analysis is an un-supervised learning technique that is widely used in the process of topic discovery from text. The research presented here proposes a novel un-supervised learning approach based on aggregation of clusterings produced by different clustering techniques. By examining and combining two different clusterings of a document collection, the aggregation aims at revealing a better structure of the data rather than imposing one that is imposed or constrained by the clustering method itself. When clusters of documents are formed, a process called topic extraction picks terms from the feature space (i.e. the vocabulary of the whole collection) to describe the topic of each cluster. It is proposed at this stage to re-compute terms weights according to the revealed cluster structure. The work further investigates the adaptive setup of the parameters required for the clustering and aggregation techniques. Finally, a topic accuracy measure is developed and used along with the F-measure to evaluate and compare the extracted topics and the clustering quality (respectively) before and after the aggregation. Experimental evaluation shows that the aggregation can successfully improve the clustering quality and the topic accuracy over individual clustering techniques.

We recently introduced the idea of solving cluster ensembles using a Weighted Shared nearest neighbors Graph (WSnnG). Preliminary experiments have shown promising results in terms of integrating different clusterings into a combined one, such that the natural cluster structure of the data can be revealed. In this paper, we further study and extend the basic WSnnG. First, we introduce the use of fixed number of nearest neighbors in order to reduce the size of the graph. Second, we use refined weights on the edges and vertices of the graph. Experiments show that it is possible to capture the similarity relationships between the data patterns on a compact refined graph. Furthermore, the quality of the combined clustering based on the proposed WSnnG surpasses the average quality of the ensemble and that of an alternative clustering combining method based on partitioning of the patterns' co-association matrix.

Traditional sketch-based image or video search systems rely on machine learning concepts as their core technology. However, in many applications, machine learning alone is impractical since videos may not be semantically annotated sufficiently, there may be a lack of suitable training data, and the search requirements of the user may frequently change for different tasks. In this work, we develop a visual analytics systems that overcomes the shortcomings of the traditional approach. We make use of a sketch-based interface to enable users to specify search requirement in a flexible manner without depending on semantic annotation. We employ active machine learning to train different analytical models for different types of search requirements. We use visualization to facilitate knowledge discovery at the different stages of visual analytics. This includes visualizing the parameter space of the trained model, visualizing the search space to support interactive browsing, visualizing candidature search results to support rapid interaction for active learning while minimizing watching videos, and visualizing aggregated information of the search results. We demonstrate the system for searching spatiotemporal attributes from sports video to identify key instances of the team and player performance.

Riassunto: In questo lavoro viene proposto un nuovo approccio, suggerito dalla Teoria dei Grafi, per ridurre, nel processo di correzione, la perdita di informazioni derivante da cancellazioni improprie dei dati, e per migliorare la qualità dell’imputazione, nel caso in cui vengano effettuati diversi passi di correzione per imputare dati con struttura gerarchica. Il nuovo approccio è stato confrontato con un approccio tradizionale usando i dati del Censimento della Popolazione del 1991; i risultati ottenuti hanno evidenziato un miglioramento consistente della qualità del processo di correzione. La metodologia proposta sarà, quindi, utilizzata per la correzione di alcune variabili rilevate con il Censimento della Popolazione del 2001.

In the paper the most recent methodological and technological advancements at ISTAT in the area of editing and imputation are described. A recently developed model-based method for localizing systematic unity measure errors and some parametric and non-parametric approaches for imputing either categorical or continuous data are illustrated. Available generalized software is also briefly presented. Recent developments concerning the DIESIS software for editing and imputing demographic census data are also discussed.

Data clustering constitutes at present a commonly used technique for extracting fuzzy system rules from experimental data. Detailed studies in the field have shown that using above-mentioned method results in significantly reduced structure of fuzzy identification system, maintaining at the same time its high modelling efficiency. In this paper a clustering algorithm, based on a kernel density gradient estimation procedure applied for fuzzy models synthesis, is presented. It consists of two stages: data elements relocation and their division into clusters. The method is automatic, unsupervised, and does not require any assumptions concerning the desired number of fuzzy rules. The results of experimental evaluation show that the algorithm under consideration achieves relatively high performance when compared to the standard techniques frequently applied in similar class of problems.

Feature extraction is an essential process in machine learning (ML). It is the process of deriving new features from the original features in order to enhance the quality or representation of the data for different reasons, such as increasing classifier efficiency and allowing higher classification accuracy. Clustering is one of the methods used in feature extraction. There are different clustering methods: soft clustering and hard clustering. Hard clustering, where each data point must be assigned to one cluster. while soft clustering, that may be a data point assigned to more than one cluster with a specific parameter "membership" value. Therefore, the type of clustering method may affect on quality of features generated using clustering. As related to using soft clustering algorithms such as Fuzzy c-means (FCM), several issues can be examined about the impact of using it on feature extraction, i. e. enhancing the quality of these features or not. Also, the effect of data circumstances that are introduced to FCM algorithm on the results. This is the problem that was adopted in this thesis. The aim is to develop a new method for feature extraction using FCM clustering to enhance classification performance, which allows for a more accurate representation of the nature of the data by using the "partial membership" features. The proposed work consists of several steps. First, an algorithm based on FCM to produce the new features would be proposed, this is because allowing points to partially belong to more than one cluster can generate useful and rich representations of features. Then, for testing the quality of these features, a classification algorithm such as K-Nearest Neighbors(KNN) algorithm would be used. At last, for comparing the results, K-means algorithm, which belongs to the hard clustering method, was used in parallel with the proposed algorithm to IX extract features independently and evaluate these features through KNN algorithm. The practical work was applied to six datasets: CICIDS2017, Breast Cancer, Sonar, Spam base, Nomao, and Diabetes. After implementing the algorithms and getting the results, we notice the superiority of the proposed method in all types of data, with some variation between data, in terms of classification performance metrics: accuracy, precision, recall, and F1 score, demonstrating its effectiveness in generating high-quality and robust features for ML tasks. CICIDS2017 dataset has achieved an accuracy of 99.10%, Breast Cancer 96.71%, Sonar 89.66%, Spam base 88.23%, Nomao 95.03%, and Diabetes 65.51%. FCM algorithm is considered better than K-Means for reasons related to the nature of the data and the characteristics of each algorithm. It allows a single point to belong to more than one cluster with different degrees and produces more expressive features. It is also suitable for cases where the class boundaries are unclear or overlap with each other.

This study explores the principal methods of data mining and their diverse applications across industries. Its purpose is to provide a comprehensive overview of key techniques-classification, clustering, association rule learning, regression, anomaly detection, text mining, and time series analysis-highlighting their theoretical foundations and real-world use cases. Methodologically, it synthesizes and analyzes existing literature, comparing algorithmic approaches such as decision trees, Naive Bayes, SVMs, k-means, DBSCAN, Apriori, FP-growth, and ARIMA models. Key findings reveal that no single method universally outperforms others; instead, optimal results often come from hybrid approaches tailored to domain-specific challenges. The study concludes that data mining remains integral for decision-making, offering predictive power, pattern discovery, and strategic insights. Its added value lies in mapping current trends, challenges, and future prospects, particularly the integration of AI and big data technologies, with implications for business intelligence, healthcare, fraud detection, and customer relationship management.

Data mining are data analysis supported unsupervised clustering algorithm is one of the quickest growing research areas because of availability of huge quantity of data analysis and extract usefully information based on new improve performance of clustering algorithm. Clustering is an unsupervised classification that's the partitioning of a data set in a set of meaningful subsets .Machine learning is based on extract and mine the invisible, meaningful data from mountain of data, hidden patterns the finding out clusters may be a supported unsupervised learning. K means is one of the best unsupervised learning strategies among all partitioning primarily based clustering strategies. The proposed algorithm is improving performance of clustering algorithm (IPCA) bases on experiment on various dataset. A proposed algorithm is minimizing error and optimization in cluster and also the effectiveness of the proposed clustering algorithm.

Wireless sensor networks (WSNs) based on the Internet of Things (IoT) are now one of the most prominent wireless sensor communication technologies. WSNs are often developed for particular applications such as monitoring or tracking in either indoor or outdoor environments, where battery power is a critical consideration. To overcome this issue, several routing approaches have been presented in recent years. Nonetheless, the extension of the network lifetime in light of the sensor capabilities remains an open subject. In this research, a CUCKOO-ANN based optimization technique is applied to obtain a more efficient and dependable energy efficient solution in IoT-WSN. The proposed method uses time constraints to minimize the distance between sources and sink with the objective of a low-cost path. Using the property of CUCKOO method for solving nonlinear problem and utilizing the ANN parallel handling capability, this method is formulated. The resented model holds significant promise si...

As network traffic gets more complex, conventional manual techniques of identifying network traffic are becoming less successful. Fraudulent activities are no longer allowed by internet auction sites like those that allow shill bidding; hence, they are insufficient. In this project, we will: develop a hybrid machine-based efficient Shill Bidding Fraud Detection (SBFD) system; recursive feature elimination (RFE) for feature selection enhances learning. The suggested model combines logistic regression, support vector machines (SVM), and In a voting ensemble structure, the Random Forest (RF) approach provides better categorization. performance and durability. The data for the network traffic of the system was preprocessed before evaluation and training. Categorical throughout the board Features were standardized; variables were label encoded. RFE is used to pick the top 10. Features were chosen from the original 41 in order to reduce computing costs and improve accuracy; the hybrid model was then trained on 80% of the data and evaluated on the remaining 20% with exceptional precision of 97. 72%, a F1 score of 0. 98, and a ROC-AUC score of more than the reference study's stated 96% accuracy and 94% precision at 0. 999. paper. Among the assessment markers confirmed were precision, recall, confusion matrix, and ROC-AUC curve. The model's great reliability in identifying both shill attackers and normal activity. The outcome was that saved for more investigation were the actual, predicted, and probability scores of the paper. Our ensemble method employs a hybrid model (SVM + Decision Tree) with an added logistic. The regression classifier showed better generalization and detecting performance.

Data mining are data analysis supported unsupervised clustering algorithm is one of the quickest growing research areas because of availability of huge quantity of data analysis and extract usefully information based on new improve performance of clustering algorithm. Clustering is an unsupervised classification that's the partitioning of a data set in a set of meaningful subsets .Machine learning is based on extract and mine the invisible, meaningful data from mountain of data, hidden patterns the finding out clusters may be a supported unsupervised learning. K means is one of the best unsupervised learning strategies among all partitioning primarily based clustering strategies. The proposed algorithm is improving performance of clustering algorithm (IPCA) bases on experiment on various dataset. A proposed algorithm is minimizing error and optimization in cluster and also the effectiveness of the proposed clustering algorithm.

Segmentation is a fundamental step in image description or classiÿcation. In recent years, several computational models have been used to implement segmentation methods but without establishing a single analytic solution. However, the intrinsic properties of neural networks make them an interesting approach, despite some measure of ine ciency. This paper presents a clustering approach for image segmentation based on a modiÿed fuzzy approach for image segmentation (ART) model. The goal of the proposed approach is to ÿnd a simple model able to instance a prototype for each cluster avoiding complex post-processing phases. Results and comparisons with other similar models presented in the literature (like self-organizing maps and original fuzzy ART) are also discussed. Qualitative and quantitative evaluations conÿrm the validity of the approach proposed.

Conversational message thread identification regards a wide spectrum of applications, ranging from social network marketing to virus propagation, digital forensics, etc. Many different approaches have been proposed in literature for the identification of conversational threads focusing on features that are strongly dependent on the dataset. In this paper, we introduce a novel method to identify threads from any type of conversational texts overcoming the limitation of previously determining specific features for each dataset. Given a pool of messages, our method extracts and maps in a three dimensional representation the semantic content, the social interactions and the timestamp; then it clusters each message into conversational threads. We extend our previous work by introducing a deep learning approach and by performing new extensive experiments and comparisons with classical learning algorithms.

Conversational message thread identification regards a wide spectrum of applications, ranging from social network marketing to virus propagation, digital forensics, etc. Many different approaches have been proposed in literature for the identification of conversational threads focusing on features that are strongly dependent on the dataset. In this paper, we introduce a novel method to identify threads from any type of conversational texts overcoming the limitation of previously determining specific features for each dataset. Given a pool of messages, our method extracts and maps in a three dimensional representation the semantic content, the social interactions and the timestamp; then it clusters each message into conversational threads. We extend our previous work by introducing a deep learning approach and by performing new extensive experiments and comparisons with classical learning algorithms.

Electric vehicles (EVs) are emerging as the future of individual mobility systems in smart cities since they reduce greenhouse gas emissions and fossil fuel dependence. However, the deepening penetration of battery EVs forecasted for the incoming years could cause significant stress on distribution networks (DNs), as well as the need to address the growing energy demand. In order to limit the negative drawbacks associated with EVs charging demand, the paper proposes a centralized approach for the EVs smart charging, and its performance are compared with the uncontrolled charging approach. An optimization framework is formulated in order to reduce both the overall peak power demand and the EVs charging cost according to the electricity prices during the day. Finally, several Monte Carlo simulations are carried out to evaluate the benefits introduced by the proposed scheduling strategy on a real case study, in terms of charging cost for EVs’ users, satisfaction of EVs charging needs, ...

Panel data, also known as longitudinal data, is collected and analysed across various research areas. This type of data consists of statistical objects that are periodically observed over time. In comparison to cross-sectional data, there are significantly fewer clustering techniques available for panel data. Therefore, the main objective of the contribution is to present a two-step clustering approach. In the first step, the panel data are transformed into a static form using a set of proposed characteristics that capture time dynamics. In the second step, the objects are clustered using conventional spatial clustering algorithms, such as K-means clustering or hierarchical partitioning. The clustering performance of this approach is then compared to that of the well-known KML method using real panel data sets. These datasets include indicators that assess the effectiveness of courts at the first instance level. Factors like digitalisation in the public sector are affecting the judiciary's efficiency during this period. The methodology implemented allowed us to categorise European countries based on the efficiency of their courts while capturing the dynamic trends. This approach is generally helpful in assessing and comparing the efficiency of public spending and evaluating the quality of public institutions, including courts.

This paper is devoted to the proposal of two classes of compromise conditional Gaussian networks for data clustering as well as to their experimental evaluation and comparison on synthetic and real-world databases. According to the reported results, the models show an ideal trade-o between eciency and eectiveness, i.e., a balance between the cost of the unsupervised model learning process and the quality of the learnt models. Moreover, the proposed models are very appealing due to their closeness to human intuition and computational advantages for the unsupervised model induction process, while preserving a rich enough modeling power.

AbstractÐThis paper introduces a novel enhancement for unsupervised learning of conditional Gaussian networks that benefits from feature selection. Our proposal is based on the assumption that, in the absence of labels reflecting the cluster membership of each case of the database, those features that exhibit low correlation with the rest of the features can be considered irrelevant for the learning process. Thus, we suggest performing this process using only the relevant features. Then, every irrelevant feature is added to the learned model to obtain an explanatory model for the original database which is our primary goal. A simple and, thus, efficient measure to assess the relevance of the features for the learning process is presented. Additionally, the form of this measure allows us to calculate a relevance threshold to automatically identify the relevant features. The experimental results reported for synthetic and real-world databases show the ability of our proposal to distinguish between relevant and irrelevant features and to accelerate learning; however, still obtaining good explanatory models for the original database.

This version may not include final proof corrections and does not include published layout or pagination. Citation for the version of the work held in 'OpenAIR@RGU': BRUZA, P. D. and SONG, D., 2003. A comparison of various approaches for using probabilistic dependencies in language modeling.

Data clustering is an approach for automatically finding classes, concepts, or groups of patterns. It also aims at representing large datasets by a few number of prototypes or clusters. It brings simplicity in modelling data and plays an important role in the process of knowledge discovery and data mining. Data mining tasks require fast and accurate partitioning of huge datasets, which may come with a variety of attributes or features. This imposes computational requirements on the clustering techniques. Swarm Intelligence (SI) has emerged that meets these requirements and has successfully been applied to a number of real world clustering problems. This paper looks into the use of Particle Swarm Optimization for cluster analysis. The effectiveness of Fuzzy C-means clustering provides enhanced performance and maintains more diversity in the swarm and allows the particles to be robust to trace the changing environment. Data structure identifying from the large scale data has become a ...

Feature selection aims to reduce dimensionality for building comprehensible learning models with good generalization performance. Feature selection algorithms are largely studied separately according to the type of learning: supervised or unsupervised. This work exploits intrinsic properties underlying supervised and unsupervised feature selection algorithms, and proposes a unified framework for feature selection based on spectral graph theory. The proposed framework is able to generate families of algorithms for both supervised and unsupervised feature selection. And we show that existing powerful algorithms such as ReliefF (supervised) and Laplacian Score (unsupervised) are special cases of the proposed framework. To the best of our knowledge, this work is the first attempt to unify supervised and unsupervised feature selection, and enable their joint study under a general framework. Experiments demonstrated the efficacy of the novel algorithms derived from the framework.

Real-world datasets commonly present high dimensional data, which means an increased amount of information. However, this does not always imply an improvement in learning technique performance. Furthermore, some features may be correlated or add unexpected noise, thereby reducing data clustering performance. This has motivated the development of feature selection methods to find the most relevant subset of features to describe data. In this work, we focus on the problem of unsupervised feature selection. The main goal is to define a method to identify the number of features to select after sorting them based on some criterion. This task is done by means of the False Nearest Neighbor technique, which is rooted in chaos theory. Results have shown that this technique gives a good approximate number of features to select. When compared to other techniques, in most of the analyzed cases, it maintains the quality of the generated partitions while selecting fewer features.

In this paper we have presented an effective hybrid genetic algorithm for solving clustering problems with multi-dimensional grid structure. The algorithm is basically a combination of Genetic Algorithm (GA) and Tabu Search (TS) so that we can efficiently utilize the stochastic search ability of GA and the hill climbing as well as the local search capabilities of TS. Such hybridization helps to enhance the capability of both the search techniques and to reduce their disadvantages. The application of TS along with GA also greatly reduces the possibility of a stochastic search process to be trapped in a local optimal solution. The proposed grid structure based clustering method is a two-step process. It starts with decomposing the data set into a finite number of grid cells. At the end of the grid partitioning process, each non-empty grid cell is considered to be a small cluster or sub-cluster. In the second step, the hybrid genetic algorithm is invoked to merge the sub-clusters hierarchically and iteratively so that the expected set of k clusters is finally obtained. The performance of this new technique has been tested on various multi-dimensional synthetic and real data sets. A comparison with related clustering and classification techniques have also been made on some data sets.

Parallel implementations of two computer vision algorithms on distributed cluster platforms are described. The rst algorithm is a square-error data clustering method whose parallel implementation is based on the well-known sequential CLUSTER program. The second algorithm is a motion parameter estimation algorithm used to determine correspondence between two images taken of the same scene. Both algorithms have been implemented and tested on cluster platforms using the PVM package. Performance measurements demonstrate that it is possible to attain good performance in terms of execution time and speedup for large-scale problems, provided that adequate memory, swap space, and I/O capacity are available at each node.

Algorithmic enhancements are described that enable large computational reduction in mean square-error data clustering. These improvements are incorporated into a parallel data-clustering tool, P-CLUSTER, designed to execute on a network of workstations. Experiments involving the unsupervised segmentation of standard texture images were performed. For some data sets, a 96 percent reduction in computation was achieved.

Cluster analysis is one of the prominent techniques in the field of data mining and k-means is one of the most well known popular and partitioned based clustering algorithms. K-means clustering algorithm is widely used in clustering. The performance of k-means algorithm will affect when clustering the continuous data. In this paper, a novel approach for performing k-means clustering on continuous data is proposed. It organizes all the continuous data sets in a sorted structure such that one can find all the data sets which are closest to a given centroid efficiently. The key institution behind this approach is calculating the distance from origin to each data point in the data set. The data sets are portioned into k-equal number of cluster with initial centroids and these are updated all at a time with closest one according to newly calculated distances from the data set. The experimental results demonstrate that proposed approach can improves the computational speed of the direct k-means algorithm in the total number of distance calculations and the overall time of computations particularly in handling continuous data.

A fundamental assumption often made in unsupervised learning is that the problem is static, i.e., the description of the classes does not change with time. However, many practical clustering tasks involve changing environments. It is hence recognized that the methods and techniques to analyze the evolving trends for changing environments are of increasing interest and importance. Although the problem of clustering numerical time-evolving data is well-explored, the problem of clustering categorical time-evolving data remains as a challenging issue. In this paper, we propose a generalized clustering framework for categorical time-evolving data, which is composed of three algorithms: a drifting-concept detecting algorithm that detects the difference between the current sliding window and the last sliding window, a data-labeling algorithm that decides the most-appropriate cluster label for each object of the current sliding window based on the clustering results of the last sliding window, and a cluster-relationship-analysis algorithm that analyzes the relationship between clustering results at different time stamps. The time-complexity analysis indicates that these proposed algorithms are effective for large datasets. Experiments on a real dataset show that the proposed framework not only accurately detects the drifting concepts but also attains clustering results of better quality. Furthermore, compared with the other framework, the proposed one needs fewer parameters, which is favorable for specific applications.

We investigate the adaptation and performance of modularity-based algorithms, designed in the scope of complex networks, to analyze the mesoscopic structure of correlation matrices. Using a multiresolution analysis, we are able to describe the structure of the data in terms of clusters at different topological levels. We demonstrate the applicability of our findings in two different scenarios: to analyze the neural connectivity of the nematode Caenorhabditis elegans and to automatically classify a typical benchmark of unsupervised clustering, the Iris dataset, with considerable success.

Context prediction is useful for energy saving and hence eco-efficient context-aware service by increasing the interval of context sensing. One way of predicting context is to recognize context patterns in an accurate manner. Traditionally, clustering method has been widely used in pattern recognition, as well as image processing and data analysis. Clustering aims to organize a collection of data items into a specific number of clusters, such that the data items within a cluster are more similar to each other than they are items in the other clusters. In this paper, a modified fuzzy possibilistic clustering algorithm is proposed based on the conventional Fuzzy Possibilistic C-means (FPCM) to obtain better quality clustering results. To show the feasibility and performance of the proposed method, numerical simulation is performed in an actual amusement park setting. The results of the numerical simulation shows that the proposed clustering algorithm gives more accurate clustering results than the FCM and FPCM methods.

Content inappropriate for children on Internet television is a serious problem in today's multimedia world. There are numerous methods which are used to control the content of the transmitted television programmes. However, these well-known methods do not solve the above mentioned problem completely. The paper presents a more effective method for automatic identification of the provider's logo based on an original image sequence analysis. The automatic identification of the provider's logo can be used to block access to video programmes of the selected providers. The method has been tested on some chosen video transmissions on-line, achieving over 98% of correct identification.

Complex systems are described with high-dimensional data that is hard to visualise. Inselberg's parallel coordinates are one representation technique for visualising high-dimensional data. Here we generalise Inselberg's approach, and use it for visualising trajectories through high dimensional state spaces. We introduce two geometric projections of parallel coordinate representations -'plan tuple plots' and 'side tuple plots' -and demonstrate a link between state space and ordinary space representations. We provide examples from many domains to illustrate use of the approach, including Cellular Automata, Random Boolean Networks, coupled logistic maps, reservoir computing, search algorithms, Turing Machines, and flocking.

The organizations of Flying ad hoc networks (FANETs) are turning into a favorable answer for various purposes situation including unmanned aerial vehicles, as metropolitan reconnaissance or search and salvage missions. Be that as it may, such organizations present different and unmistakable correspondence issues. As a result, there are a few exploration contemplates zeroed in on examining their exhibition through recreation. Accurately demonstrating versatility is urgent in this unique circumstance and albeit numerous portability models are now profit ready to repeat the conduct of versatile hubs in a specially appointed organization, the majority of these models can't be utilized to dependably reproduce the movement of automated ethereal automobiles. In this paper, we list the current portability models and give direction to comprehend whether they could be really received relying upon the particular flying specially appointed organization application situations, while talking about their points of interest and burdens.

Clustering is widely used to explore and understand large collections of data. In this thesis, we introduce LIMBO, a scalable hierarchical categorical clustering algorithm based on the Information Bottleneck (IB) framework for quantifying the relevant information preserved when clustering. As a hierarchical algorithm, LIMBO can produce clusterings of different sizes in a single execution. We also define a distance measure for categorical tuples and values of a specific attribute. Within this framework, we define a heuristic for discovering candidate values for the number of meaningful clusters. Next, we consider the problem of database design, which has been characterized as a process of arriving at a design that minimizes redundancy. Redundancy is measured with respect to a prescribed model for the data (a set of constraints). We consider the problem of doing database redesign when the prescribed model is unknown or incomplete. Specifically, we consider the problem of finding structural clues in a data instance, which may contain errors, missing values, and duplicate records. We propose a set of tools based on LIMBO for finding structural summaries that are useful in characterizing the information content of the data. We study the use of these summaries in ranking functional dependencies based on their data redundancy. We also consider a different application of LIMBO, that of clustering software artifacts. The majority of previous algorithms for this problem utilize structural information in order to decompose large software systems. Other approaches using non-structural iniii formation, such as file names or ownership information, have also demonstrated merit. We present an approach that combines structural and non-structural information in an integrated fashion. We apply LIMBO to two large software systems, and the results indicate that this approach produces valid and useful clusterings. Finally, we present a set of weighting schemes that specify objective assignments of importance to the values of a data set. We use well established weighting schemes from information retrieval, web search and data clustering to assess the importance of whole attributes and individual values.

Fuzzy clustering algorithm is one of the data mining methods that is applied in different fields. According to the fuzzy clustering algorithm, each object is allocated to the clusters regarding its percentage of belonging to each of the clusters. Finding the cluster centers is one of the main objectives of the clustering and it is possible to apply the swarm intelligence methods in order to accurately find the cluster centers. The swarm intelligence algorithms have separately been applied for clustering and they received the optimized solution. In order to solve the problems of the fuzzy clustering algorithm, the combined method based on the firefly algorithm and the particle swarm optimization algorithm is applied. In order to determine the validity, the suggested method is tested on four standard data collections received from the valid site of UCI. The simulation results shows that the combination of two algorithms do the clustering more accurately than applying each of them sepa...

The huge size of multimedia data requires for efficient data classification and organization in providing effective multimedia data manipulation. Those valuable data must be captured and stored for potential purposes. One of the main problems in Multimedia Information System (MIS) is the management of multimedia data. As a consequence, multimedia data management has emerged as an important research area for querying, retrieving, inserting and updating of these vast multimedia data. This research considers the rough set theory technique to organize and categorize the multimedia data. Rough set theory method is useful for exploring multimedia data and simplicity to construct multimedia data classification. Classification will help to improve the performance of multimedia data retrieving and organizing process.

This study examined the level of community satisfaction with the operational performance of the 1st Provincial Mobile Force Company (PMFC) in Basilan concerning its anti-criminality campaign. Using a descriptive research design, data were collected through a researcher-made survey questionnaire distributed among community sector-respondents in multiple municipalities under the jurisdiction of the 1st PMFC. Findings indicated varying levels of satisfaction across municipalities, with Lamitan City and Albarka Municipality reporting "Very Satisfactory" ratings, while areas such as Akbar, Tuburan, Tipo-Tipo, and Ungkaya Pukan showed "Satisfactory" ratings. The results suggested that while the 1st PMFC has demonstrated effective crime prevention and response strategies in some areas, further improvements are needed to enhance law enforcement efficiency, particularly in municipalities with lower satisfaction ratings. The study highlighted the importance of continuous evaluation, community engagement, and strategic police interventions to ensure public safety and maintain community trust in law enforcement agencies.

Clustering is a widely used technique in data mining application for discovering patterns in underlying data. Most traditional clustering algorithms are limited in handling datasets that contain categorical attributes. However, datasets with categorical types of attributes are common in real life data mining problem. For these data sets, no inherent distance measure, like the Euclidean distance, would work to compute the distance between two categorical objects. In this article, we have described two algorithms based on genetic algorithm and simulated annealing in the field of crisp and fuzzy domain. The performance of the proposed algorithms has been compared with that of different well known categorical data clustering algorithms in crisp and fuzzy domain and demonstrated for a variety of artificial and real life categorical data sets. Also statistical significance tests have been performed to establish the superiority of the proposed algorithms.

Knowledge discovery in multi-dimensional data is a challenging problem in engineering design. For example, in trade space exploration of large design data sets, designers need to select a subset of data of interest and examine data from different data dimensions and within data clusters at different granularities. This exploration is a process that demands both humans, who can heuristically decide what data to explore and how best to explore it, and computers, which can quickly extract features that may be of interest in the data. Thus, to support this process of knowledge discovery, we need tools that can go beyond traditional computeroriented optimisation approaches and support advanced designer-centred trade space exploration and data interaction. This paper is an effort to address this need. In particular, we propose the interactive multiscalenested clustering and aggregation framework to support trade space exploration of multi-dimensional data common to design optimisation. A system prototype of this framework is implemented to allow users to visually examine large design data sets through interactive data clustering, aggregation, and visualisation. The paper also presents an evaluation study involving morphing wing design using this prototype system.

We develop a new method to measure neutron star parameters and derive constraints on the equation of state of dense matter by fitting the frequencies of simultaneous Quasi Periodic Oscillation modes observed in the X-ray flux of accreting neutron stars in low mass X-ray binaries. To this aim we calculate the fundamental frequencies of geodesic motion around rotating neutron stars based on an accurate general-relativistic approximation for their external spacetime. Once the fundamental frequencies are related to the observed frequencies through a QPO model, they can be fit to the data to obtain estimates of the three parameters describing the spacetime, namely the neutron star mass, angular momentum and quadrupole moment. From these parameters we derive information on the neutron star structure and equation of state. We present a proof of principle of our method applied to pairs of kHz QPO frequencies observed from three systems (4U1608-52, 4U0614+09 and 4U1728-34). We identify the kHz QPOs with the azimuthal and the periastron precession frequencies of matter orbiting the neutron star, and via our Bayesian inference technique we derive constraints on the neutrons stars' masses and radii. This method is applicable to other geodesic-frequency-based QPO models.

The clustering algorithms have evolved over the last decade. With the continuous success of natural inspired algorithms in solving many engineering problems, it is imperative to scrutinize the success of these methods applied to data clustering. These naturally inspired algorithms are mainly stochastic search and optimization techniques, guided by the principles of collective behavior and self-organization of insect swarms. The parameters setting of the ant colony clustering algorithms determine the behavior of each ant and are critical for fast convergence to near optimal solutions of clustering task. This inspired us to explore techniques for automatically learning the optimal parameters for a given clustering task. We devised and implemented a hybrid Ant-Colony clustering algorithm, which uses particle swarm optimization algorithm in the early stages to 'breed' a population of ants possessing near optimal behavioral parameter settings for a given problem. This hybrid algorithm converges rapidly for nearly optimal parameters that maximize the ant-colony clustering behavior.