Search

We introduce a semi-supervised space adjustment framework in this paper. In the introduced framework, the dataset contains two subsets: (a) training data subset (space-one data (SOD)) and (b) testing data subset (space-two data (STD)). Our semi-supervised space adjustment framework learns under three assumptions: (I) it is assumed that all data points in the SOD are labeled, and only a minority of the data points in the STD are labeled (we call the labeled space-two data as LSTD), (II) the size of LSTD is very small comparing to the size of SOD, and (III) it is also assumed that the data of SOD and the data of STD have different distributions. We denote the unlabeled space-two data by ULSTD, which is equal to STD - LSTD. The aim is to map the training data, i.e., the data from the training labeled data subset and those from LSTD (note that all labeled data are considered to be training data, i.e., SOD ∪ LSTD) into a shared space (ShS). The mapped SOD, ULSTD, and LSTD into ShS are named MSOD, MULSTD, and MLSTD, respectively. The proposed method does the mentioned mapping in such a way that structures of the data points in SOD and MSOD, in STD and MSTD, in ULSTD and MULSTD, and in LSTD and MLSTD are the same. In the proposed method, the mapping is proposed to be done by a principal component analysis transformation on kernelized data. In the proposed method, it is tried to find a mapping that (a) can maintain the neighbors of data points after the mapping and (b) can take advantage of the class labels that are known in STD during transformation. After that, we represent and formulate the problem of finding the optimal mapping into a non-linear objective function. To solve it, we transform it into a semidefinite programming (SDP) problem. We solve the optimization problem with an SDP solver. The examinations indicate the superiority of the learners trained in the data mapped by the proposed approach to the learners trained in the data mapped by the state of the art methods. [ABSTRACT FROM AUTHOR]

Existing fuzzy clustering ensemble approaches do not consider dependability. This causes those methods to be fragile in dealing with unsuitable basic partitions. While many ensemble clustering approaches are recently introduced for improvement of the quality of the partitioning, but lack of a median partition based consensus function that considers more participate reliable clusters, remains unsolved problem. Dealing with the mentioned problem, an innovative weighting fuzzy cluster ensemble framework is proposed according to cluster dependability approximation. For combining the fuzzy clusters, a fuzzy co-association matrix is extracted in a weighted manner out of initial fuzzy clusters according to their dependabilities. The suggested objective function is a constrained nonlinear objective function and we solve it by sparse sequential quadratic programming (SSQP). Experimentations indicate our method can outperform modern clustering ensemble approaches. [ABSTRACT FROM AUTHOR]

Optimization is a process that is followed to improve an objective function. The aim of the optimization is to improve the solution of a problem developed by a scientist or an engineer. In any optimization algorithm, the initial conditions have a big effect on quality of the final found solution. One of the new optimization algorithms is See-See Partridge Chicks Optimization (SSPCO) algorithm, which is modeled upon the behavior of the partridge chicks. Chaotic behaviors are frequent in the natural phenomena. Indeed, many real world behaviors which resemble random, are chaotic. As these behaviors are highly sensitive to the initial conditions, they show such really random-like behaviors that can't be predicted (while they are not random at all). In this paper, we have implemented a chaotic SSPCO (CSSPSO) algorithm. We have used two different chaotic functions and we have lastly compared the results with the state of the art algorithms. The results indicate that chaos has a positive effect and the CSSPCO algorithm outperforms the state of the art algorithms. [ABSTRACT FROM AUTHOR]

Abundant data of the patients is recorded within the health care system. During data mining process, we can achieve useful knowledge and hidden patterns within the data and consequently we will discover the meaningful knowledge. The discovered knowledge can be used by physicians and managers of health care to improve the quality of their services and to reduce the number of their medical errors. Since by the usage of a single data mining algorithm, it is difficult to diagnose or predict diseases, therefore in this research, we take a combination of the advantages of some algorithms in order to achieve better results in terms of efficiency. Most of standard learning algorithms have been designed for balanced data (the data with the same frequency of samples in each class), where the cost of wrong classification is the same within all classes. These algorithms cannot properly represent data distribution characteristics when datasets are imbalanced. In some cases, the cost of wrong classification can be very high in a sample of a special class, such as wrongly misclassifying cancerous individuals or patients as healthy ones. In this article, it is tried to present a fast and efficient way to learn from imbalanced data. This method is more suitable for learning from the imbalanced data having very little data in class of minority. Experiments show that the proposed method has more efficiency compared to traditional simple algorithms of machine learning, as well as several special-to-imbalanced-data learning algorithms. In addition, this method has lower computational complexity and faster implementation time. [ABSTRACT FROM AUTHOR]

Nature is a huge source of inspiration for solving difficult and complex problems in science. Nature-based problem solvers always find an optimal solution (whether local or global) to any given problem; so they are sometimes considered as black-box problem solvers. Meta-heuristic algorithms that are inspired by nature (through imitating the nature) have opened a new approach to solving optimization problems. In the past decades, numerous research efforts have been focused in this specific area. In this paper, an optimization algorithm inspired by the nature has been introduced which is modeled from the behavior of the chicks of a type of bird called see-see partridge. We name the algorithm SSPCO. SSPCO optimization algorithm is an algorithm similar to the algorithm of particle swarm optimization (PSO) that the motion equation and the variables' velocity are modeled from the behavior of this type of bird. Simulation of the algorithm was done with MATLAB software and the results of the 6 primary known algorithms and 8 advanced ones were compared with 14 unconstrained continuous functions. The results indicate that the proposed algorithm is an efficient algorithm comparing the others. [ABSTRACT FROM AUTHOR]

Data pre-processing is one of the crucial phases of data mining that enhances the efficiency of data mining techniques. One of the most important operations performed on data pre-processing is missing values imputation in incomplete datasets. This research presents a new imputation technique using K-means and samples weighting mechanism based on Grey relation (KWGI). The Grey-based K-means algorithm applicable to all samples of incomplete datasets clusters the similar samples, then an appropriate kernel function generates appropriate weights based on the Grey relation. The missing values estimation of the incomplete samples is done based on the weighted mean to reduce the impact of outlier and vague samples. In both clustering and imputation steps, a penalty mechanism has been considered to reduce the similarity of ambiguous samples with a high number of missing values, and consequently, increase the accuracy of clustering and imputation. The KWGI method has been applied on nine natural datasets with eight state-of-the-art and commonly used methods, namely CMIWD, KNNI, HotDeck, MeanI, KmeanI, RKmeanI, ICKmeanI, and FKMI. The imputation results are evaluated by the Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) criteria. In this study, the missing values are generated at two levels, namely sample and value, and the results are discussed in a wide range of missingness from low rate to high rate. Experimental results of the t-test show that the proposed method performs significantly better than all the other compared methods. [ABSTRACT FROM AUTHOR]

With the massive amount of data available on the Internet, many users face problems when they need access to the information and goods they need. Adapting information to user needs has become a complex and time-consuming process. Recommender systems are powerful tools for guiding users in electronic settings to information, services, and goods of interest. With the ability to identify users and predict their preferences, recommender systems can extract information likely to be of interest to users from vast amounts of data and save time and energy by providing them with relevant recommendations. Therefore, in this paper, a framework for a recommender system is proposed, which consists of four main phases. In the first phase, the preprocessing operation is performed. The purpose of this phase is to create an ontology model for products based on the convolutional neural network method. Data collection operations are carried out in the second phase. The purpose of this phase is to receive and store user information based on behavior and different criteria. The third phase is related to creating an ontology. The purpose of this phase is to create an ontology model for user behavior using OWL libraries. Finally, in the last phase, finding similarities and presented a proposal. The purpose of this phase is the necessary calculations of the proposed framework in the similarity finding section using ontology models. Simulation results show that the proposed framework measures MAE (more than 10 and 15%), RMSE (more than 12 and 16%), precision (less than 10 and 13%), recall (more than 25 and 28%), F1-score (higher than 25 and 30%) and Score of User (higher than 15 and 17%) outperform two related approaches, namely CF and CF+ Ontology, respectively. [ABSTRACT FROM AUTHOR]

Clustering as a major task in data mining is responsible for discovering hidden patterns in unlabeled datasets. Finding the best clustering is also considered as one of the most challenging problems in data mining. Due to the problem complexity and the weaknesses of primary clustering algorithm, a large part of research has been directed toward ensemble clustering methods. Ensemble clustering aggregates a pool of base clusterings and produces an output clustering that is also named consensus clustering. The consensus clustering is usually better clustering than the output clusterings of the basic clustering algorithms. However, lack of quality in base clusterings makes their consensus clustering weak. In spite of some researches in selection of a subset of high quality base clusterings based on a clustering assessment metric, cluster-level selection has been always ignored. In this paper, a new clustering ensemble framework has been proposed based on cluster-level weighting. The certainty amount that the given ensemble has about a cluster is considered as the reliability of that cluster. The certainty amount that the given ensemble has about a cluster is computed by the accretion amount of that cluster by the ensemble. Then by selecting the best clusters and assigning a weight to each selected cluster based on its reliability, the final ensemble is created. After that, the paper proposes cluster-level weighting co-association matrix instead of traditional co-association matrix. Then, two consensus functions have been introduced and used for production of the consensus partition. The proposed framework completely overshadows the state-of-the-art clustering ensemble methods experimentally. [ABSTRACT FROM AUTHOR]

In clustering ensemble, it is desired to combine several clustering outputs in order to create better results than the output results of the basic individual clustering methods in terms of consistency, robustness and performance. In this research, we want to present a clustering ensemble method with a new aggregation function. The proposed method is named Robust Clustering Ensemble based on Iterative Fusion of Base Clusters (RCEIFBC). This method takes into account the two similarity criteria: (a) one of them is the cluster-cluster similarity and (b) the other one is the object-cluster similarity. The proposed method has two steps and has been done on the binary cluster representation of the given ensemble. Indeed, before doing any step, the primary partitions are converted into a binary cluster representation where the primary ensemble has been broken into a number of primary binary clusters. The first step is to combine the primary binary clusters with the highest cluster-cluster similarity. This phase will be replicated as long as our desired candidate clusters are ready. The second step is to improve the merged clusters by assigning the data points to the merged clusters. The performance and robustness of the proposed method have been evaluated over different machine learning datasets. The experimentation indicates the effectiveness of the proposed method comparing to the state-of-the-art clustering methods in terms of performance and robustness. [ABSTRACT FROM AUTHOR]

The increasing pollution caused by conventional cars and the problems caused by the use of fossil fuels have drawn the attention of researchers and manufacturers to the design of cars that use clean fuels. Electric vehicles connected to the network have a significant impact on reducing environmental pollution and transportation costs, especially in big cities. The cost of supplying loads to subscribers in the distribution network also includes generation and transmission costs. These costs are directly related to the intelligence of the distribution network and the total amount of energy of electric vehicles. The contribution of each generation unit and each transmission line must be calculated to determine the generation and transmission costs. In this research, in order to maximize the profit of the parking lot owner, improve voltage drop and load factor, a comprehensive framework for optimal energy management in a parking lot is presented, which can provide a method to control the charging of electric vehicles, in addition to meeting the needs of their owners, only as a series of controllable loads that they need to receive electrical energy to charge their batteries. In the next step, considering the inherent characteristic of electric cars, i.e. having a battery, and looking at them as a series of storage resources that can return the electric energy in their battery to the grid if necessary, a method to simultaneously control their charging and discharging is provided. In the final step of the paper, it is assumed that hydrogen storage systems will also enter the circuit, and thus, a comprehensive method for energy management is proposed. Finally, the linearized model of demand response and the proposed scheme along with the modeling of hydrogen storage and electric vehicles are considered to be part of contribution to improve the operation and economic situation of the network. • In practice, the high cost that battery depreciation imposes on car owners when using the V2G feature makes the use of this feature not economically justified. • A comprehensive framework for optimal energy management in a parking lot with the aim of maximizing the profit of the parking lot owner, improving the voltage drop and network power factor is provided. • The proposed method for energy management in the presence of hydrogenous storage systems was applied in a sample parking lot and how it works. • Considering the hydrogenous storage systems, while significantly increasing the profit of the parking lot owner, it also significantly reduces the power received from the network during the peak hours of the network. [ABSTRACT FROM AUTHOR]

With the creation of competitive environments, such as electricity market, it is expected that energy networks and active consumers such as energy hubs participate in the market to promote their economic situation. So, the article proposes the optimal involvement of energy networks and hubs in energy markets in two wholesale and retail designs based on the energy management system at the same time. The proposed scheme is expressed as two-objective optimization. The first objective is to minimize the cost of different types of energy in electricity-gas-thermal networks. In another objective function, the cost of energy (which is the difference between the energy purchase cost and energy sale income) of energy hubs in the retail market is minimized. The scheme is bound to optimal power flow equations of the mentioned networks and operating model of power sources and active loads. Then, the Pareto optimization mixed with the sum of weighted functions helps extract an optimal compromise solution on the basis of fuzzy decision-making. Finally, the scheme is applied to a test system, and the obtained numerical results confirm that energy hubs are improved financially, and economic and operation conditions of the electricity-gas-thermal networks are enhanced simultaneously. So, significant profit can be achieved for EHs in the retail energy market. The economic situation of the networks enhances up to roughly 10% compared with that of power flow studies. Also, operating situation of the networks enhances by about 12% to 53% compared with a case without EHs. [ABSTRACT FROM AUTHOR]

Software developers want to meet the requirements of customers in next versions. Choosing which set of requirements can be done according to cost and time is an NP-hard problem known as Next Release Problem (NRP). In this article, a multi objective evolutionary algorithm (MOEA) framework is proposed to solve NRP. The framework applies the non-repetitive population, integrates solutions and external repository. Furthermore, a novel approach is implemented to satisfy the constraints of the problem. In this framework, six evolutionary algorithms are implemented and using seven quality indicators, the achieved results of that algorithms are compared with the original versions of same algorithms. Through using HV (the ratio of the region covered by Pareto Front) and NDS (the number of solutions in the Pareto Front) metrics, the effects of the proposed algorithms are compared with other works' results. The efficacy of the proposed MOEA framework is measured using three real world datasets. The gained results represent that the implemented algorithms perform better than other related algorithms previously published. [ABSTRACT FROM AUTHOR]

In this paper, the optimal scheduling of energy grids and networked energy hubs based on their participation in the day-ahead energy wholesale and retail markets is presented. The problem is formulated as a bilevel model. Its upper level minimizes the expected energy cost of electricity, gas, and heating grids, especially in the form of private distribution companies in the mentioned markets, in the first objective function, and it minimizes the expected energy loss of these networks in the second objective function. This problem is constrained by linearized optimal power flow equations. The lower-level formulation minimizes the expected energy cost of hubs (equal to the difference between sell and purchase of energy) as an objective function in the retail market. Constraints of this model are the operation formulation of sources and active loads and the flexibility limit of hubs. The unscented transformation approach models the uncertainties of load, renewable power, energy price, and energy demand of mobile storage. Then, the Karush–Kuhn–Tucker approach and Pareto optimization technique based on ε-constraint are adopted to extract the single-level single-objective formulation. Finally, obtained results verify the capability of the present method in improving the economic status of hubs and the economic and operation situation of the mentioned networks simultaneously so that the proposed scheme by managing the power of energy hubs compared with power flow studies has been able to reduce operating costs by 8%, reduce energy losses by 10%, and improve voltage profile and temperature by 36% and 30%. [ABSTRACT FROM AUTHOR]

• Optimal management (technical and economic) of generation expansion planning. • Improve the performance of generation expansion planning based on demand response programs. • Reduce operating costs of generation expansion planning based on models of probability distribution functions. • Increase the reliability and security of distribution networks based on the adaptive particle swarm optimization algorithm. Generation expansion planning in the power system is of particular importance. In traditional systems, investment in the generation expansion was made by the electricity company, but with the restructuring in the electricity industry, the owners of different parts of the system submit their proposals to the independent system operator and the independent system operator chooses the optimal design. Slowly increasing energy production from renewable sources can pose challenges for the grid. Increasing the penetration of renewable resources due to uncertainty in their production can reduce network reliability and thus increase system costs. The investigation on generation expansion planning is a multifaceted issue (technical and economic) that has been analyzed in various aspects in recent years. In this study, a multidimensional structure of generation expansion planning based on increasing the penetration level of distributed generation resources (renewable and non-renewable) as well as the application of load management and demand response algorithms is proposed. The proposed model is scheduled based on two levels of primary and secondary development. In the primary, the development of generation and transmission based on large-scale power plants as well as solar and wind farms are presented. In the secondary, in order to reduce the power fluctuations caused by the distributed generation's units, non-stochastic power generation units such as micro turbines, gas turbines and combined heat and power have been utilized. To overcome the difficulties in solving the problem of hybrid and non-convergent mixed-integer problem, the adaptive particle swarm optimization has been hired. The simulation results indicate that in the second scenario, where the development of the generation expansion planning is based on the integration of distributed generation resources and power plants, it is more cost-effective. In addition to, these simulation results represent the accuracy of the proposed probabilistic method in planning of dynamic generation systems in order to estimate the probability density function and the optimal output variables in multi-objective techno-economic planning. [ABSTRACT FROM AUTHOR]

In order to improve the performance of a clustering on a data set, a number of primary partitions are generated and stored in an ensemble and their aggregated consensus partition is used as their clustering. It is widely accepted that the consensus partition outperforms the primary partitions. In this paper, an ensemble clustering method called multi-level consensus clustering (MLCC) is proposed. To construct the MLCC, a cluster–cluster similarity matrix which is achieved by an innovative similarity metric is first generated. The mentioned cluster–cluster similarity matrix is based on a multi-level similarity metric. In fact, it can be computed in a new defined multi-level space. Then, a point–point similarity matrix which is boosted using the mentioned cluster–cluster similarity matrix is generated. The new consensus function applies an average linkage hierarchical clusterer algorithm on the mentioned point–point similarity matrix to make consensus partition. MLCC is better than traditional clustering ensembles and simple versions of clustering ensembles on traditional cluster–cluster similarity matrix. Its computational cost is not very bad too. Accuracy and robustness of the proposed method are compared with those of the modern clustering algorithms through the experimental tests. Also, time analysis is presented in the experimental results. [ABSTRACT FROM AUTHOR]

One of the main goals of SDN is to achieve the desirable operation of a highly secure and reliable network in terms of energy distribution. Hence, this paper presents the probabilistic planning of distributed generations (DGs) and switched capacitive bank (SCB) constrained to the securable-reliable operation (SRO) strategy in reconfigurable SDN. The scheme is formulated as a four-objective optimization problem to simultaneously model the economic, operation, reliability, and security indices of SDN. So that the first to fourth objective minimize the total planning cost and expected operation cost, voltage deviation function (VDF), expected energy not-supplied (EENS), and voltage security index (VSI). It is limited to the AC optimal power flow (AC-OPF) equations, planning and operation model of DG and SCB, formulation of demand response program (DRP), SDN reconfiguration constraints, reliability and voltage security limits. Then, the single-objective model is obtained by using the ε constraint-based Pareto optimization technique. Additionally, the non-parametric probabilistic method models uncertainties of load, energy price, renewable DGs (RDGs) power, and network equipment availability. The strategy is as mixed-integer nonlinear programming (MINLP), where hybrid Krill Herd Optimization (KHO) and Crow Search Algorithm (CSA) achieves a reliable optimal solution. The following provides the planning of different sources based on the SRO strategy with simultaneous formulation of various economic and technical indices considering non-parametric modeling of uncertainties and solving the problem using the mentioned hybrid algorithm presented the contribution in this paper. By implementing this scheme on a 69-bus SDN, the numerical results confirm capability of this scheme in improving the economic, operation, reliability, and security situation of SDN compared to power flow studies. Where it can enhance the mentioned indices up to 16%, 36%, 99%, and 9% taking into account optimal planning and operation of different sources and the DRP compared to the corresponding values obtained by power flow studies. • Probabilistic planning modeling of active and reactive power sources constrained to secure-reliable operation in a reconfigurable smart distribution network. • Simultaneous modeling of economic, operation, reliability, and voltage security indices of SDN by adopting a four-objective optimization for the proposed plan. • Using a hybrid KHO+CSA evolutionary algorithm to achieve a reliable optimal solution with low standard deviation in response. • Utilizing non-parametric probabilistic planning to model uncertainties of load, energy cost, RDG capacity, and availability of SDN equipment. Flowchart of the proposed solution method [Display omitted] [ABSTRACT FROM AUTHOR]

Due to the pattern of growth for electricity consumption, there is a need for developing power networks and transmission lines. The power transmission capacity of lines is limited due to a host of factors. Thus, these lines need series and parallel compensations to reduce losses, increase efficiency, and promote system security. In this paper, flexible alternating current transmission system (FACTS) devices including static VAR compensators (SVC) as parallel compensators, thyristor-controlled series compensation (TCSC) as a series compensator, and high-voltage direct current (HVDC) bonding are modeled. In addition, comprehensive modeling of the simultaneous application of these three devices for load flow is performed, and the effects of these types of compensations are compared. The obtained comprehensive model was implemented on MATLAB software using the Newton–Raphson method on two 9-bus WSCC and 5-bus test system. In this case, the calculation speed and convergence were reduced when compared to applying devices individually due to the increase in equations and the addition of new terms to the load flow equations. Furthermore, more losses were observed in this model, which can probably be improved using an optimal power flow and optimal placement of the devices in the network. [ABSTRACT FROM AUTHOR]

Job scheduling in Hadoop has been thus far investigated in several studies. However, some challenges including minimum share (min-share), heterogeneous cluster, execution time estimation, and scheduling program size facing Hadoop clusters have received less attention. Accordingly, one of the most important algorithms with regard to min-share is that presented by Facebook Inc., i.e., FAIR scheduler, based on its own needs, in which an equal min-share has been considered for users. In this article, an attempt has been made to make the proposed method superior to existing methods through automation and configuration, performance optimization, fairness and data locality. A high-level architectural model is designed. Then a scheduler is defined on this architectural model. The provided scheduler contains four components. Three components schedule jobs and one component distributes the data for each job among the nodes. The given scheduler will be capable of being executed on heterogeneous Hadoop clusters and running jobs in parallel, in which disparate min-shares can be assigned to each job or user. Moreover, an approach is presented for each problem associated with min-share, cluster heterogeneity, execution time estimation, and scheduler program size. These approaches can be also utilized on its own to improve the performance of other scheduling algorithms. The scheduler presented in this paper showed acceptable performance compared with First-In, First-Out (FIFO), and FAIR schedulers. [ABSTRACT FROM AUTHOR]

A significant amount of research in the field of job scheduling is carried out in Hadoop. However, there is still need for research to overcome some challenges regarding scheduling jobs in Hadoop clusters. There are various factors affecting the performance of scheduling policies like data volume (storage), data source format (different data), speed (data rate), security and privacy, cost, connection and data sharing. To reach a better utilization of resources and managing big data, scheduling policies have been designed. In this paper, an algorithm has been presented that can run on heterogeneous Hadoop clusters and runs job in parallel. This algorithm first distributes data based on the performance of the nodes and then schedules the jobs according to their cost of execution and decreases the cost of executing the jobs. The presented algorithm offers better performance in terms of execution time, cost and locality compared to FIFO and Fair schedulers. [ABSTRACT FROM AUTHOR]

Model checking is one of the most efficient techniques in software system verification. However, state space explosion is a big challenge while using this technique to check different properties like safety ones. In this situation, one can search the state space to find a reachable state in which the safety property is violated. Hence, reachability checking can be done instead of checking safety property. However, checking reachability in the worst case may cause state space explosion again. To handle this problem, our idea is based on generating a small model consistent with the main model. Then by exploring the state space entirely, we search it to find the goal states. After finding the goal states, we label the paths which start from the initial state and leading to a goal state. Then using the ensemble classification technique, the necessary knowledge is extracted from these paths to intelligently explore the state space of the bigger model. Ensemble machine learning technique uses Boosting method along with decision trees. It follows sampling techniques by replacement. This method generates k predictive models after sampling k times. Finally, it uses a voting mechanism to predict the labels of the final path. Our proposed approach is implemented in GROOVE, which is an open source toolset for designing and model checking graph transformation systems. Our experiments show a significant improvement in terms of both speed and memory usage. In average, our approach consumes nearly 42% fewer memory than other approaches. Also, it generates witnesses nearly 20% shorter than others, in average. [ABSTRACT FROM AUTHOR]

After observing the ensemble success in supervised learning (such as classification), it was extended into unsupervised learning. Therefore, cluster ensemble, which merges multiple basic data partitions or clusters (called as ensemble pool) into an ordinarily better clustering solution usually named as consensus partition, emerged. Any cluster ensemble method tries to optimize a particular criterion during extracting the consensus partition out of the ensemble pool. But traditional cluster ensembles consider all the pool members with the equal importance in making the consensus partition; that is to say that each basic partition or cluster participates in the cluster ensemble algorithm equivalently. Indeed, they ignore to consider any ensemble member according to its importance. But it is obvious that some clusters with more quality deserve more emphasis and some clusters with less quality deserve less emphasis during generating consensus partition. This paper proposes (a) a metric to evaluate quality of any arbitrary cluster, (b) a mechanism to project the computed quality of a cluster into a meaningful weight value, and (c) an approach to apply the weight values of the basic clusters in the cluster ensemble process. Experimental results conducted on a number of real‐world standard datasets indicate that the proposed method outperforms the state of the art methods. [ABSTRACT FROM AUTHOR]

Since complexity of computer systems is growing increasingly, assuring flawless operation of these systems has become more difficult. Therefore, it is important that these systems whether software or hardware are executed as expected. Consequently, verifying system before implementation at model level is necessary. Model checking is a formal technique for validating the system automatically which decides whether the finite state system satisfies temporal property by scanning the whole state space or not. One of the most important problems in model checking is state space explosion of models which results in memory shortage in generation of all states. Therefore, this paper presents a method which employs machine learning techniques without exploring the whole state space to predict temporal properties of trajectories in systems based on graph transmission system. the proposed method is implemented in Groove; results indicate desirable accuracy and speed of this method compared to other methods. [ABSTRACT FROM AUTHOR]

The paper has proposed a linear unsupervised transfer learning (LUTL). Therefore, a cost function has been introduced. In the cost function of the proposed LUTL, the aim is to minimize the difference between the distribution of the transformed source domain (SD) data and the distribution of the target domain (TD) data. In the proposed cost function, it is also targeted to preserve the local structures of the untransformed SD data. Three mechanisms have been proposed for the preservation of local structures in the untransformed SD data: (1) minimization of the distances between the data pairs that are similar to each other in the untransformed SD data, (2) preservation of the clusters emerged in the untransformed SD data and finally (3) their combination. The optimization problem has emerged as a nonlinear one. Two techniques have been introduced to obtain an approximation of the optimal weight matrix. Each technique guarantees to reach a local optimum, but no one guarantees to reach the global solution. While the first method is an iterative one, the second is a relaxed version of the optimization problem. The paper shows also experimentally that the proposed techniques overshadow the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Bagging and Boosting are two main ensemble approaches consolidating the decisions of several hypotheses. The diversity of the ensemble members is considered to be a significant element to obtain generalization error. Here, an inventive method called EBAGTS (ensemble-based artificially generated training samples) is proposed to generate ensembles. It manipulates training examples in three ways in order to build various hypotheses straightforwardly: drawing a sub-sample from training set, reducing/raising error-prone training instances, and reducing/raising local instances around error-prone regions. The proposed method is a straightforward, generic framework utilizing any base classifier as its ensemble members to assemble a powerfully built combinational classifier. Decision-tree classifier and multilayer perceptron classifier as some basic classifiers have been employed in the experiments to indicate the proposed method accomplish higher predictive accuracy compared to meta-learning algorithms like Boosting and Bagging. Furthermore, EBAGTS outperforms Boosting more impressively as the training data set gets broader. It is illustrated that EBAGTS can fulfill better performance comparing to the state of the art. [ABSTRACT FROM AUTHOR]