Search

Over the years the number of research publications per year is growing exponentially. Finding research papers of quality from the massive literature of relevant articles is a challenging and time-consuming task. The approaches in the latest literature address citation recommendation by utilizing large bibliographic information and use machine learning and deep learning methods for the task. These techniques clearly require a large amount of training data as well as machines with high processing power. To overcome these issues, we propose a novel method by modifying the popular hyperlink induced topic search (HITS), a web page ranking algorithm, as citation recommendation using hyperlink induced topic search (CR-HITS) that works on a directed and weighted heterogeneous bibliographic network containing diverse types of nodes and edges. We define effective scoring schemes for nodes and edges based on basic bibliographic information like citations of papers, number of publications of an author, etc. Given a few seed papers, the citation recommendation algorithm CR-HITS is run on small neighborhoods of the seed papers and hence the time taken by the execution is very small to yield the final recommendations. To the best of our knowledge, HITS has been used for the first time for the citation recommendation problem. We perform extensive experimentation on DBLP (version-11) and ACM (version-9) datasets and compare the results with many baseline methods in terms of MAP, MRR, and recall@N measures. The performance of the proposed algorithms is superior with respect to the MAP metric and matches the second best for the other two metrics. Since the top two algorithms use deep learning methods and use much larger bibliographic information including abstracts of the papers, we claim that our approach utilizes very low resources, yet yields recommendations that are very close to the top recommendations. [ABSTRACT FROM AUTHOR]

In recent years, Sine Cosine Algorithm (SCA) is a kind of meta-heuristic optimization algorithm with simple structure, simple parameters and trigonometric function principle. It has been proved that it has good competitiveness among the existing optimization algorithms. However, the single mechanism of SCA leads to its insufficient utilization of the information of the whole population, insufficient ability to jump out of local optima and poor performance at solving complex objective function. Therefore, this paper introduces social learning strategy (SL) and elite opposition-based learning (EOBL) strategy to improve SCA, and proposes novel algorithm: enhancing Sine Cosine Algorithm based on elite opposition-based learning and social learning (ESLSCA). Social learning strategy takes full advantage of information from the entire population. The elite opposition-based learning strategy provides a possibility for the algorithm to jump out of local optima and increases the diversity of the population. To demonstrate the performance of ESLSCA, this paper uses 22 well-known benchmark test functions and CEC2019 test function set to evaluate ESLSCA. The comparisons show that the proposed ESLSCA has better performance than the standard SCA and it is very competitive among other excellent optimization algorithms. [ABSTRACT FROM AUTHOR]

Cloud data centers face various challenges, such as high energy consumption, environmental impact, and quality of service (QoS) requirements. Dynamic virtual machine (VM) consolidation is an effective approach to address these challenges, but it is a complex optimization problem that involves trade-offs between energy efficiency and QoS satisfaction. Moreover, the workload patterns in cloud data centers are often non-stationary and unpredictable, which makes it difficult to model them. In this paper, we propose a new method for dynamic VM consolidation that optimizes both energy efficiency and QoS objectives. Our approach is based on Markov chains and the artificial feeding birds (AFB) algorithm. Markov chains are used to model the resource utilization of each individual VM and PM based on the changes that happen in workload data. AFB algorithm is a metaheuristic optimization technique that mimics the behavior of birds in nature. We modify the AFB algorithm to suit the characteristics of the VM placement problem and to provide QoS-aware and energy-efficient solutions. Our approach also employs an online step detection method to capture variations in workload patterns. Furthermore, we introduce a new policy for VM selection from overloaded hosts, which considers the abrupt changes in the utilization processes of the VMs. The proposed algorithms are evaluated extensively using the CloudSim Toolkit with real workload data. The proposed system outperforms evaluation policies in multiple metrics, including energy consumption, SLA violations, and other essential metrics. [ABSTRACT FROM AUTHOR]

Virtual Machine Placement (VMP) plays a significant role in improving efficiency of Cloud Data Center (CDC). With the dramatic increase in the use of cloud computing, it seems necessary to apply effective algorithms to reduce the power consumption of CDC. VMP is known as a NP-Hard problem that cannot be solved by deterministic algorithms in polynomial time. In this paper, an algorithm named Combinated Random Best First Fit (CRBFF) is proposed with the aim of increasing the Quality of Service (QoS), in which Virtual Machines (VMs) are optimally placed on heterogeneous Physical Machines (PMs). The effectiveness of CRBFF is evaluated by different metrics on Google Compute Engine (GCE), Amazon Web Service Elastic Compute Cloud (AWS EC2) and Microsoft Azure scenarios and the results show that CRBFF performs better than other common algorithms. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) is constantly evolving. The variety of IoT applications has caused new demands to emerge on users' part and competition between computing service providers. On the one hand, an IoT application may exhibit several important criteria, such as deadline and runtime simultaneously, and it is confronted with resource limitations and high energy consumption on the other hand. This has turned to adopting a computing environment and scheduling as a fundamental challenge. To resolve the issue, IoT applications are considered in this paper as a workflow composed of a series of interdependent tasks. The tasks in the same workflow (at the same level) are subject to priorities and deadlines for execution, making the problem far more complex and closer to the real world. In this paper, a hybrid Particle Swarm Optimization and Simulated Annealing algorithm (PSO–SA) is used for prioritizing tasks and improving fitness function. Our proposed method managed the task allocation and optimized energy consumption and makespan at the fog-cloud environment nodes. The simulation results indicated that the PSO–SA enhanced energy and makespan by 5% and 9% respectively on average compared with the baseline algorithm (IKH-EFT). [ABSTRACT FROM AUTHOR]

The K-means algorithm has been successfully applied to many complex network analysis problems. However, this method is sensitive to how the first cluster centers are chosen. It is possible to minimize superfluous runs by choosing the first cluster center in advance because each run produces a unique set of results. To overcome this issue, an algorithm for Community Detection based on Transitive Closure (CoDTC) has been introduced. In this algorithm, the initial cluster center is provided by degree centrality and T-transitive closure. The algorithm initializes with the calculation of the similarity relation matrix. Then, to avoid the limitation of sparse problems in complex network analysis, we offer the idea of transitive closure on weighted networks to solve the sparsity issue. This notion is based on imposing a t-norm inequality on the connection weights and providing a method to compute them. Finally, based on T-transitive closure, new cluster centers are calculated iteratively to avoid random selection of cluster centers. In this paper, we demonstrate the efficacy of the CoDTC approach on a diverse range of real and artificial networks, encompassing both big and small communities. [ABSTRACT FROM AUTHOR]

Classification is a widely used supervised learning technique that enables models to discover the relationship between a set of features and a specified label using available data. Its applications span various fields such as engineering, telecommunication, astronomy, and medicine. In this paper, we propose a novel rule-based classifier called RUCIB (RUle-based Classifier Inspired by BRADO), which draws inspiration from the socio-inspired swarm intelligence algorithm known as BRADO. RUCIB introduces two key aspects: the ability to accommodate multiple values for features within a rule and the capability to explore all data features simultaneously. To evaluate the performance of RUCIB, we conducted experiments using ten databases sourced from the UCI machine learning database repository. In terms of classification accuracy, we compared RUCIB to ten well-known classifiers. Our results demonstrate that, on average, RUCIB outperforms Naive Bayes, SVM, PART, Hoeffding Tree, C4.5, ID3, Random Forest, CORER, CN2, and RACER by 9.32%, 8.97%, 7.58%, 7.4%, 7.34%, 7.34%, 7.22%, 5.06%, 5.01%, and 1.92%, respectively. [ABSTRACT FROM AUTHOR]

It is axiomatic that influence maximization is one of the major issues of the Internet today. In this paper a novel specialized metaheuristic algorithm is proposed to efficiently deal with it. The purpose of influence maximization is to select a subset of seed nodes in the network in order to influence other nodes maximally. The greedy methods present good results for influence maximization, but these algorithms involve very high computational time. Unlike greedy algorithms, meta-heuristic algorithms have acceptable efficiency. The ICO algorithm is just recently proposed as an ensign of meta-heuristic algorithms to solve continuous optimization problems in 2022. In this paper, a discrete version of ICO, called DICO, is proposed. The contribution of the proposed method is to present a new intelligent operator based on the nodes' degree and logistic mapping for initialization of solutions. In addition, novel approaches are proposed to reproduce each parent discretely. Another innovation of the proposed method is to present a new local search operator based on the network topology. In this operator, a tabu list is suggested to decrease the selection probability of nodes influenced by those nodes selected previously. The proposed method is evaluated on 6 real- world networks, and it is compared to the state-of-the-art and conventional methods. The evaluation results prove that the proposed method has the higher influence than most of other methods. Also, it has an acceptable performance in terms of computational time making it prominent in comparison with existing methods. [ABSTRACT FROM AUTHOR]

This paper first presents a high-efficient and accurate coupled meshless algorithm for solving the multi-dimensional Gross–Pitaevskii equation (GPE) in unbounded domain, which is implemented on CUDA-program-based two-GPUs cards. The proposed novel high-performance scheme (RDFPM-PML-GPU) is mainly motived by the items below: (a) a reduced-dimensional finite pointset method (RDFPM) is first presented to solve the 2D/3D spatial derivatives in GPE, which has lower calculated amount than the traditional FPM (TFPM) for the derivatives; (b) the perfectly matched layer (PML) technique is adopted to treat the absorbing boundary conditions (ABCs) which is used for the infinite exterior region, and the time-splitting technique is resorted to reduce the computing complexity in PML; (c) a fast parallel algorithm based on CUDA-program is proposed to accelerate the computation in the proposed meshless scheme with local matrix on two GPUs. The numerical convergent rate and advantages of the proposed meshless scheme are demonstrated by solving two examples, which include the comparisons between the proposed RDFPM and TFPM, the merit of easily implemented local refinement point distribution in meshless method, and the merit of PML-ABCs over the zero Dirichlet boundary treatment for unbounded domain. Meanwhile, the high efficiency of the proposed GPU-based parallelization algorithm is tested and discussed by simulating 3D examples, which shows that the speed-up rate is about 500-times of using two-GPUs over a single CPU. Finally, the proposed RDFPM-PML-GPU method is used to predict the long-time evolution of quantum vortex in 2D/3D GPEs describing Bose–Einstein condensates. All the numerical tests show the high-performance and flexible application of the proposed parallel meshless algorithm. [ABSTRACT FROM AUTHOR]

Workflow applications are a popular tool used by scientists to model and run applications on cloud computing. With the rise of multi-core parallel systems and cloud computing high-performance computing is becoming more accessible to scientists, but it also presents new challenges. One pressing issue is the energy consumption of these systems, both for environmental and financial reasons. To minimize energy usage in cloud computing, methods such as energy-efficient scheduling are gaining attention. However, current scheduling solutions have limitations, or failing to address the problem as a multi-objective optimization balancing performance and energy. This problem is considered complex and NP-complete. Many researchers have attempted to resolve it using heuristic and meta-heuristic methods. Although these methods may not always provide optimal results, they are still a subject of active research. This paper proposes a new multi-objective optimization algorithm that combines the Heterogeneous Earliest Finish Time and BAT algorithm to optimize multiple conflicting objectives for energy, cost, makespan, and resource utilization. The results are verified using the Analysis of Variance statistical tool, and the proposed algorithm is shown to be superior to existing contemporary algorithms. [ABSTRACT FROM AUTHOR]

Machine learning techniques aim to mimic the human ability to automatically learn how to perform tasks through training examples. They have proven capable of tasks such as prediction, learning and adaptation based on experience and can be used in virtually any scientific application, ranging from biomedical, robotic, to business decision applications, and others. However, the lack of domain knowledge for a particular application can make feature extraction ineffective or even unattainable. Furthermore, even in the presence of pre-processed datasets, the iterative process of optimizing Machine Learning parameters, which do not translate from one domain to another, maybe difficult for inexperienced practitioners. To address these issues, we present in this paper a Vectorized Automated ML Pre-processIng and post-pRocEssing framework, approximately named (VAMPIRE), which implements feature extraction algorithms capable of converting large time-series recordings into datasets. Also, it introduces a new concept, the Activation Engine, which is attached to the output of a Multi Layer Perceptron and extracts the optimal threshold to apply binary classification. Moreover, a tree-based algorithm is used to achieve multi-class classification using the Activation Engine. Furthermore, the internet of things gives rise to new applications such as remote sensing and communications, so consequently applying Machine Learning to improve operation accuracy, latency, and reliability is beneficial in such systems. Therefore, all classifications in this paper were performed on the edge in order to reach high accuracy with limited resources. Moreover, forecasts were applied on three unrelated biomedical datasets, and on two other pre-processed urban and activity detection datasets. Features were extracted when required, and training and testing were performed on the Raspberry Pi remotely, where high accuracy and inference speed were achieved in every experiment. Additionally, the board remained competitive in terms of power consumption when compared with a laptop which was optimized using a Graphical Processing Unit. [ABSTRACT FROM AUTHOR]

The evolution of cloud computing results in the development of numerous huge data-centers. These data-centers use large quantities of electrical energy, which in turn results in high operating cost and high emissions of CO 2 . The grounds for this excessive usage of energy is the inefficient usage of data-center resources (hosts). The resources in the data-center can be in either peak or idle or sleep state. It has been experimentally proven that hosts in the idle state consumes more energy than hosts in sleep state. Through an efficient and optimal VM placement strategy, an energy efficient resource utilization can be achieved, thereby reducing the number of idle hosts in the data-center. This paper introduces two energy efficient VM placement algorithms based on bin packing heuristics considering the physical machine's energy efficiency, Energy Efficient VM Placement (EEVMP) and Modified Energy Efficient VM Placement (MEEVMP), that can reduce the overall energy usage in the data-center. The EEVMP algorithm when compared with the default VM placement algorithm Power-Aware Best-Fit Decreasing algorithm (PABFD) of CloudSim, it reduces the energy consumption by 53%, average SLA violation by 3.5% and number of VM migrations by 64.47%. Further, we have performed MEEVMP algorithm where we achieve a reduction in energy consumption by 54.24%, average SLA violation by 4.39% and number of VM migrations by 67.713 % as compared to PABFD. [ABSTRACT FROM AUTHOR]

In this paper, we propose a novel indoor localization system in a multi-indoor environment using cloud computing. Prior studies show that there are always concerns about how to avoid signal occlusion and interference in the single indoor environment. However, we find some general rules to support our system being immune to interference generated by occlusion in the multi-indoor environment. A convenient way is measured to deploy Bluetooth low energy devices, which mainly collect large information to assist localization. A neural network-based classification is proposed to improve localization accuracy, compared with several algorithms and their performance comparison is discussed. We also design a distributed data storage structure and establish a platform considering the storage load with Redis. Our real experimental validation shows that our system will meet the four aspects of performance requirements, which are higher accuracy, less power consumption, and increased levels of system magnitude and deployment efficiency. [ABSTRACT FROM AUTHOR]

There are several public cloud service providers (CSPs) across the globe supplying a variety of application, platform, middleware, database, and infrastructure services. The brewing challenge before any cloud user is how to be sure about the trustworthiness of a service being offered by various CSPs. With the overwhelming usage of cloud services by individuals and organizations, this problem has acquired a lot of attention these days. Precisely arriving and articulating that this particular service from a specific cloud CSP is trustworthy is becoming complicated because of many moving variables in this whole phenomenon. There are varying parameters and indicators for decisively proving that a particular service is trustworthy or not. In order to meet up this crucial challenge and concern being widely expressed by cloud customers, researchers and cloud professionals across the world have unearthed a few viable mechanisms. There are fresh algorithms, techniques, and tools to enable cloud users towards easily and quickly selecting trustworthy services. This paper digs deeper and dwells at length about the intrinsic challenges being associated with the provenance of the trustworthiness factor. The paper also presents a machine learning approach to compute the trust factor of CSPs and it is found that the methodology gains advantage when compared to similar works. [ABSTRACT FROM AUTHOR]

In anonymous distributed systems, processes are indistinguishable because they have no identity and execute the same algorithm. Currently, anonymous systems are receiving a lot of attention mainly because they preserve privacy, which is an important property when we want to avoid impersonation attacks. On the other hand, Consensus is a fundamental problem in distributed computing. It is well-known that Consensus cannot be deterministically solved in pure asynchronous anonymous systems if processes can crash (the so-called crash-stop failure model). This impossibility holds even if message losses never occur in transmission. Failure detectors are an elegant and powerful abstraction for achieving deterministic Consensus in asynchronous distributed systems. A failure detector is a distributed object that gives the processes information about crashed processes. Failure detectors have attracted so much attention in the crash-stop failure model because they provide a totally independent abstraction. Ω is the weakest failure detector to solve Consensus in classic asynchronous systems when a majority of processes never crash, and A Ω ′ is its implementable version for anonymous systems. As far as we know, there is a lack of works in the literature which tackle Consensus in anonymous asynchronous systems where crashed process can recover (the so-called crash-recovery failure model) and also assuming errors in transmission operations (the so-called omission failure model). Extending failure models in the system allows us to design more realistic systems and solve more practical security problems (i.e., fair exchange and the secure multiparty computation). We present, in this paper, an algorithm to solve Consensus using A Ω ′ in anonymous asynchronous systems under the crash-recovery and omission failure models. Another important contribution of this paper is a communication-efficient and latency-efficient implementation of A Ω ′ for these new failure models. [ABSTRACT FROM AUTHOR]

The structural similarity based link prediction algorithms mainly exploit the information of network topology, such as links and nodes, to predict the potential links in complex networks. Among these algorithms, the local information similarity based algorithms have attracted the extensive attentions from the majority of researchers due to their low complexity and general applicability. The algorithms mainly exploit the attributes of common neighbors on the second-order transmission paths to predict the connection probability between the unconnected nodes, but ignore the structure attributes of endpoints. The structure attributes of an endpoint can be quantified as its influence resources, which make an important contribution to link prediction. To heighten the performances of local information based algorithms, this paper exploits the different structure attributes of endpoints to express the influence resources, and explores the contributions of the different endpoint attributes to local information algorithms. Extensive simulations on 12 real benchmark datasets show that, in most cases, the node degree expressing the influence resource makes the greatest contribution to improve the performances of the local information algorithms. Specifically, DCN, DAA and DRA algorithm possess the best prediction performances in 9, 5 and 7 datasets, respectively. Furthermore, compared with 6 mainstream algorithms, DRA as the best improved algorithm shows the optimal prediction performances in 8 datasets. [ABSTRACT FROM AUTHOR]

Overlapping community detection with low computation is one of the fundamental issues and challenges in large-scale complex network analysis. Detecting a community in a network means discovering a cluster of network nodes so that the density of edges between them is high. The existing methods use entire structure information or subgraphs with a fixed size to detect dense communities. Therefore, they are not efficient and accurate for large-scale networks. In this paper, the authors introduce an overlapping community detection algorithm that gradually improves density estimation by expanding the size of subgraphs and gathering information during the search process for finding communities. It is an efficient algorithm with low computational complexity for complex networks with one hundred thousand to several millions of nodes. Experimental results on synthetic and real-world networks with several hundred to four million nodes validate the performance assessment of the proposed method. [ABSTRACT FROM AUTHOR]

This paper presents a hybrid approach based Binary Artificial Bee Colony (BABC) and Pareto Dominance strategy for scheduling workflow applications considering different Quality of Services (QoS) requirements in cloud computing. The main purpose is to schedule a given application onto the available machines in the cloud environment with minimum makespan (i.e. schedule length) and processing cost while maximizing resource utilization without violating Service Level Agreement (SLA) among users and cloud providers. The proposed approach is called Enhanced Binary Artificial Bee Colony based Pareto Front (EBABC-PF). Our proposed approach starts by listing the tasks according to priority defined by Heterogeneous Earliest Finish Time (HEFT) algorithm, then gets an initial solution by applying Greedy Randomized Adaptive Search Procedure (GRASP) and finally schedules tasks onto machines by applying Enhanced Binary Artificial Bee Colony (BABC). Further, several modifications are considered with BABC to improve the local searching process by applying circular shift operator then mutation operator on the food sources of the population considering the improvement rate. The proposed approach is simulated and implemented in the WorkflowSim which extends the existing CloudSim tool. The performance of the proposed approach is compared with Heterogeneous Earliest Finish Time (HEFT) algorithm, Deadline Heterogeneous Earliest Finish Time (DHEFT), Non-dominated Sort Genetic Algorithm (NSGA-II) and standard Binary Artificial Bee Colony (BABC) algorithm using different sizes of tasks and various benchmark workflows. The results clearly demonstrate the efficiency of the proposed approach in terms of makespan, processing cost and resources utilization. [ABSTRACT FROM AUTHOR]

In this paper, we propose the first self-stabilizing c-wave algorithm for broadcasting large size messages from multiple initiator processes such that every process receives at least one of these messages broadcast by the initiators. The proposed algorithm first splits each message into a sequence of message shares and then broadcasts the message shares in a pipelined manner in order. We then present an improved maximally concurrent self-stabilizing c-wave algorithm called dynamic clusteringc-wavealgorithm that constructs dynamic broadcast trees by allowing a process to dynamically change its parent to a neighbouring process with the most message shares to implement a maximally concurrent broadcasting of large size messages. The improved algorithm is maximally concurrent in the sense that a process is allowed to receive a message share from any neighbouring process with a larger sequence of message shares than its current parent from which it receives message shares in the broadcast trees constructed by the initiators when available. We implemented and experimentally evaluated the c-wave algorithm to compare its message propagation delay to those of the m and the k-wave algorithms. Simulation results show that the proposed algorithm significantly reduces the broadcast propagation delays compared to that of the m and the k-wave algorithms. The improvement is more evident over the m-wave algorithm. Solutions to global-snapshots, distributed broadcast and various synchronization problems can be solved efficiently using c-wave algorithms. In addition, the proposed algorithms have applications in blockchain, mobile wireless sensor networks, VANETs and military communication networks. [ABSTRACT FROM AUTHOR]

For many years, job scheduling in cloud computing has been researched to improve and optimize the environment. Although many researchers have worked on the issue of job scheduling, however, a comprehensive approach still misses out on various fronts like consideration of multi objective functions, handling the problem of local minima, and best resource utilization. An attempt has been made in the paper to present a reliable and comprehensive scheduling approach based on the meta-heuristic for the cloud computing environment. The proposed algorithm imitates the behavior of Rock Hyrax optimization for scheduling the jobs in a dynamic and heterogeneous cloud environment by considering the quality of service parameters like makespan time and energy consumption of data centers. The result establishes the claim that the proposal presented in this paper can schedule jobs in a dynamic environment on the virtual machine by keeping energy consumption low. The proposal is implemented through an experimental setup in the CloudSim environment and considered for variable jobs. The proposed algorithm for scheduling in the cloud environment is evaluated both qualitatively and quantitatively by considering both jobs and virtual machines statically and dynamically. The proposed algorithm is also compared with the prevalent approaches proposed in the past and shows better results. Our results indicate that the proposed meta-heuristic algorithm based on Rock Hyrax has lowered the makespan time by 5–15% and reduces energy consumption by 4–12%. [ABSTRACT FROM AUTHOR]

The case study of plant intelligence inspired novel non-swarm intelligence (NSI) algorithms, namely Venus Flytrap Optimization and Bladder-Worts Suction, concentrated in this paper. These algorithms devised on the prey-hunting mechanisms of the Venus Flytrap (Dionaea Muscipula) and BladderWorts (Utricularia) plants, respectively. A comparative view of these algorithms is discussed. The main-support criterion is the major characteristic of these approaches. The benefits of this main-support criterion and their performances are evidenced with a case study of extracting the highly correlated maximal local patterns in gene expression data through biclustering. The NSI algorithms are proposed for biclustering gene expression data in this paper. The results are compared with existing optimization techniques like PSO and GA, and biclustering approaches like Cheng and Church, OPSM, BiMax, and Plaid approaches. This analysis evidenced the performance of NSI algorithms can yield optimal maximal local patterns with high correlation. Further, various real-time research applications of NSI approaches are also discussed. [ABSTRACT FROM AUTHOR]

Resource management is the process of task scheduling and resource provisioning to provide requirements of cloud users. Since cloud resources are often heterogeneous, task scheduling and resource provisioning are major challenges in this area. Various methods have been introduced to improve resource utilization and thus increase the efficiency of cloud computing. Existing methods can be divided into several categories, including mathematical and statistical methods, heuristic- and meta-heuristic-based methods, and machine-learning-based methods. Since the resource management problem is NP-complete, several optimization methods have been also exploited in this area. Coral reefs algorithm is an evolutionary method that has showed appropriate convergence and response time for some problems, and thus is used in this paper to combine with reinforcement learning to improve efficiency of resource management in cloud environments. The proposed method of this paper consists of two phases. The initial allocation of resources to ready-to-perform tasks is done using the coral reefs algorithm in the first phase. The tasks are considered as corals and the resources are considered reefs in this method. The second phase utilizes reinforcement learning to avoid falling into the local optima and to make optimal use of resources using a long-term approach. The proposed model of this paper, called MO-CRAML, introduces a new hybrid algorithm for improving utilization and load balancing of cloud resources using the combination of coral reefs optimization algorithm and reinforcement learning. The results of the experiments show that the proposed algorithm has better performance in cloud resource utilization and load balancing in comparison with some other important methods of the literature. [ABSTRACT FROM AUTHOR]

The rapid growing number of web services has posed new challenges for service composition computing. How to combine services that meet the needs of users in the least amount of time from a huge number of candidate services is a hot topic of research today. As a meta-heuristic algorithm for solving optimization problems, salp swarm algorithm (SSA) has been widely applied to case scenarios in different fields due to its simple structure and high performance. However, QoS-aware service composition is a discrete problem and existing methods are not suitable for it. Therefore, in this paper, we propose an improved SSA integrating chaotic mapping method for QoS service composition selection, named CSSA. Through the randomness and ergodicity of chaos, reducing the possibility of falling into local optimum and strengthening the exploitation capability of the algorithm. In addition, a fuzzy continuous neighborhood search method is used to enhance the local search capability of the algorithm which makes the discrete space of service composition in a way similar to continuous space. Finally, two well-known datasets are used to verify the effectiveness of CSSA compared to three advanced algorithms and original SSA. The test results demonstrate that CSSA has significant advantages and it also has satisfactory performance in large scale scenarios. [ABSTRACT FROM AUTHOR]

Today, the Internet of Things (IoT) is extending due to a wide range of applications and services. The variety of devices connected to the internet, the discussion of security on these networks is a challenging issue. Security includes diverse aspects such as botnets. Botnets are a set of devices such as smartphones, computers, and others polluted by a program. This program, which is a bot herder, performs many deleterious operations and leads to various anomalies in the network. Identifying botnets due to their unique complexity is one of the main challenges in IoT security. In this paper, we propose a model for identifying botnets in the internet of things. The proposed method is based on selecting features using the modified League Championship Algorithm (LCA) and constructing the model using artificial neural networks. Feature selection speeds up the learning process and increases the resolution of botnets. The proposed method is simulated using MATLAB. The results reveal that the proposed method can make a better detection method than other schemes, and our modified version selects an optimal subset of features. As a result, it is an efficient model. [ABSTRACT FROM AUTHOR]

While k-anonymous algorithms can effectively protect users' private location information, the problem of selecting an appropriate location in the anonymous area to construct the k-anonymous area remains a significant one. When selecting real users from the surrounding area to co-construct anonymous regions, it is easy to cause the leakage of user location information. Moreover, using false addresses to construct a region requires calculating the probability of location queries, which increases the computational complexity. In this paper, an all-dummy k-anonymous algorithm based on location offset is proposed to construct anonymous regions. This algorithm randomly selects k−1 locations and real users in the selected anonymous compose an anonymous group at first. Subsequently, these coordinates are centered on migration, generating multiple dummy addresses of each location migration, such that the dummy address distance is greater than the radius of the user's query, with the dummy address location information used for the location server queries. Through experimental verification, compared with the circle-based dummy address generation algorithm and the random k-anonymous algorithm, the all-dummy k-anonymous algorithm is found to achieve an entropy value and tracking success rate closer to the optimal k-anonymous algorithm without increasing the communication cost. [ABSTRACT FROM AUTHOR]

Cloud computing technology provides shared computing which can be accessed over the Internet. When cloud data centers are flooded by end-users, how to efficiently manage virtual machines to balance both economical cost and ensure QoS becomes a mandatory work to service providers. Virtual machine migration feature brings a plenty of benefits to stakeholders such as cost, energy, performance, stability, availability. However, stakeholders' objectives are usually conflict with each other. Furthermore, the optimal resource allocation problem in cloud infrastructure is usually NP-Hard or NP-Complete class. In this paper, the virtual migration problem is formulated by applying the game theory to ensure both load balance and resource utilization. The virtual machine migration algorithm, named V2PQL, is proposed based on Markov decision process and Q-learning algorithm. The results of the simulation demonstrate the efficiency of our proposal which are divided into training phase and extraction phase. The proposed V2PQL algorithm has been benchmarked to the Round-Robin, inverse Ant System, Max–Min Ant System, and Ant System algorithms in order to highlight its strength and feasibility in extraction phase. [ABSTRACT FROM AUTHOR]

In recent years, with the increasing volume of databases, the removal of redundant features has become an essential thing in classification. A smaller subset of features is selected using feature selection algorithm. One of the famous algorithms of feature selection methods is the crow search algorithm (CSA). This algorithm's popularity can be mentioned in the algorithm's implementation and process and the impressive results compared to the previous algorithms. Despite all these benefits, this algorithm suffers from problems such as unbalanced global and local search. It is also stuck in local optimization due to the search approach's inadequacy. In this paper, a new algorithm based on CSA is introduced. In order to overcome the shortcoming, four fundamental changes have been made to CSA. (i) The algorithm uses the concept of dynamic awareness probability to solve the balance between exploitation and exploration. Then, a new approach is introduced for each part of the search that improves crows' search performance both (ii) locally and (iii) globally. Also, as the last change, (iv) the concept of chaos is used to increase the algorithm's convergence rate. The proposed method has been tested and compared with ten well-known algorithms in this field on the same datasets and has performed on average 20% better in the feature reduction index and 2.5% in the fitness index, while has a lower performance in accuracy by only 1.5%. Practical results show that the algorithm changes have provided attractive results compared to other algorithms in this field in the mentioned metrics. [ABSTRACT FROM AUTHOR]

Recently, the concept of cloud computing has been extended towards the network edge. Devices near the network edge, called fog nodes, offer computing capabilities with low latency to nearby end devices. In the resulting fog computing paradigm (also called edge computing), application components can be deployed to a distributed infrastructure, comprising both cloud data centers and fog nodes. The decision which infrastructure nodes should host which application components has a large impact on important system parameters like performance and energy consumption. Several algorithms have been proposed to find a good placement of applications on a fog infrastructure. In most cases, the proposed algorithms were evaluated experimentally by the respective authors. In the absence of a theoretical analysis, a thorough and systematic empirical evaluation is of key importance for being able to make sound conclusions about the suitability of the algorithms. The aim of this paper is to survey how application placement algorithms for fog computing are evaluated in the literature. In particular, we identify good and bad practices that should be utilized respectively avoided when evaluating such algorithms. [ABSTRACT FROM AUTHOR]

Uniform reliable broadcast (URB) is an important abstraction in distributed systems, offering delivery guarantee when spreading messages between processes. Informally, URB guarantees that if a process (correct or not) delivers a message m, then all correct processes deliver m. This abstraction has been extensively investigated in distributed systems where all processes have unique identifiers. Furthermore, the majority of papers in the literature usually assume that the communication channels of the system are reliable, which is not always the case in real systems. In this paper, the URB abstraction is investigated in anonymous asynchronous message passing distributed systems with process crash failures and fair lossy channels. For that, we assume the availability of a random number generator that generates unique global values with very high probability. Firstly, a simple URB algorithm is given assuming a majority of correct processes. Then, we prove the impossibility of solving URB without a majority of correct processes if no failure detector is used. Subsequently, a new failure detector class A Θ is proposed, which can be used to implement URB with any number of correct processes. However, the two previous URB algorithms are non-quiescent because every correct process, to offset the loss of messages caused by fair lossy channels, has to broadcast all URB _ delivered messages forever. Hence, a perfect anonymous failure detector A P ∗ is proposed, together with A Θ , to make the URB algorithm quiescent. Finally, we discuss an alternative failure detector A Θ P ∗ , which combines the properties of A Θ and A P ∗ . [ABSTRACT FROM AUTHOR]

This paper concentrates on a parallel acceleration method of optimizing Gaussian hyper-parameters with the maximum likelihood estimation. In the process of optimizing the hyper-parameters, many calculations of the kernel matrix inversion will be operated. With an increase of the kernel matrix scale, the high computation burden will be generated. In order to improve the calculating efficiency, we introduce a decomposing and iterative (DI) algorithm. This algorithm divides the large-scale kernel matrix into four blocks and solves the matrix inversion with constant iterations. Due to the independency of the calculations of the sub-matrix blocks, it is quite suitable to put the sub-matrix blocks computation in graphics processing unit. Hence, the parallel decomposing and iterative (DIP) algorithm is introduced. The inverted pendulum and ball-plate system experiments are carried out to confirm the effectiveness of the DI and DIP algorithms. Based on the simulation results, the proposed DI and DIP algorithms shed light on real engineering application in the future. This paper also provides a practical and feasible approach to accelerate the optimization of hyper-parameters with maximum likelihood estimation. [ABSTRACT FROM AUTHOR]

In real life, some complex problems are multi-objective optimization problems. Most of the existing studies have focused on how to obtain the optimal solutions distributed on the whole Pareto-optimal frontier. However, in some fields such as industrial production, the decision-makers of enterprises usually care about what specific measures can maximize the comprehensive benefits of enterprises. Due to this kind of realistic demands, we prefer to find a small part of the optimal solutions according to the preference information suggested by the decision-makers rather than obtain all of the Pareto-optimal solutions. However, almost all of the existing methods only repeat calculation when they meet the scenario where the user-preferences change over time. To address the multi-objective optimization problem under the scenario with dynamic user-preferences information, we propose a MOEA/D-DPRE (multi-objective optimization algorithm based on dynamic preference information) algorithm in this paper, and its framework is inspired by the MOEA/D-PRE (decomposition user-preference multi-objective evolutionary) algorithm. We analyze the four position relations between the distribution region of the old preference weight vectors (old preference region), and we also present the distribution region of the new preference weight vectors (new preference region) and propose the different strategy to the different case respectively. When the preference information changes, the MOEA/D-DPRE can converge to the region of new interest by responding to the change of preference and the historical information. Experimental results show that the proposed method is better than the compared method in convergence speed and distribution under the scenario with dynamic user-preferences information. [ABSTRACT FROM AUTHOR]

The online social network has become an integral part of our day to life and serves as an excellent platform for sharing ideas, opinions, and products. Influence maximization (IM) is a widely studied topic in the area of social network analysis. The objective of IM is to find influential nodes that can disseminate information to a larger extent in the network. Many local and global centrality measures are proposed to rank the nodes based on their spreading capability with certain limitations. Many proposed algorithms locate the spreaders sharing overlapping regions or are closely placed, which may cause interference in spreading. In this paper, based on the notion of maximum coverage of the information and minimum interference in spreading, we propose a novel semi-local algorithm named as modified degree centrality with exclusion ratio to identify influential nodes from diverse locations in the network. We use modified degree centrality by considering neighbours upto 2-hops and introduce the novel idea of exclusion ratio to ensure minimum overlapping between regions influenced by the chosen spreader nodes. Extensive experimental validation using classical information diffusion model demonstrates that the proposed method delivers better results in comparison to many popular contemporary methods of influence maximization. [ABSTRACT FROM AUTHOR]

The detection of community structure plays an important role in understanding the properties and characteristics of complex networks. The label propagation algorithm (LPA) emerges as a popular community detection method, due to its simplicity and low computational cost. Nonetheless, the LPA is not without its limitations so that the Semi Synchronous Constrained Label Propagation Algorithm (SSCLPA) is a modified LPA that implements various constraints to ameliorate the stability of the LPA. Aside from giving accurate and deterministic detection, it can avoid trivial detection. In this paper the SSCLPA is extended into weighted and directed networks, so that nodes which fulfill certain conditions are updated separately at the end of the algorithm. Furthermore, some modifications are performed on the propagation processes in the SSCLPA. These new features and modifications improve the time efficiency of the SSCLPA with only marginal loss in the quality of the detection. Our proposed method is tested and compared to the other community detection methods in various benchmark and real-world networks. The results showed that the proposed method is a well-balanced method with features that takes into account the stability, quality and time efficiency of the detection. [ABSTRACT FROM AUTHOR]

Community detection is regarded as a significant research domain in social network analysis. Thanks to its merits, including linear-time complexity, performance and simplicity, the label propagation algorithm (LPA) has attracted a lot of researchers' interests in recent years. However, regarding the label propagation process, it has some drawbacks such as uncertainty and randomness behavior which may negatively impact the stability and accuracy of community detection. In this paper, a simple and fast method is proposed to overcome the LPA's drawbacks. We have proposed a novel method to improve the certainty and stability of the label propagation algorithm based on node ranking in a social network, called the CSLPR algorithm. After performing a proposed local method for ranking nodes, the label propagation is started from the nodes with lowest rank. In the first and second iterations t < = 2 of the propagation, if the number of maximum label frequency in neighboring of a node be equal, the Adamic/Adar index is used for selecting the appropriate label. For the other iterations t > 2 , a new criterion, known as label strength, is applied to select the label with the highest strength of a node. In real-world social networks, communities are established around important nodes. Therefore, when a low-significance node joins the networks, it receives the label of a community C i that has the most connection to this community. The proposed method is evaluated on several real-world and artificial networks. The results reveal that the proposed method can accurately detect communities with higher performance, certainty, and stability in comparison to other methods. [ABSTRACT FROM AUTHOR]

The different communications of users in social networks play a key role in effect to each other. The effect is important when they can achieve their goals through different communications. Studying the effect of specific users on other users has been modeled on the influence maximization problem on social networks. To solve this problem, different algorithms have been proposed that each of which has attempted to improve the influence spread and running time than other algorithms. Due to the lack of a review of the meta-heuristic algorithms for the influence maximization problem so far, in this paper, we first perform a comprehensive categorize of the presented algorithms for this problem. Then according to the efficient results and significant progress of the meta-heuristic algorithms over the last few years, we describe the comparison, advantages, and disadvantages of these algorithms. [ABSTRACT FROM AUTHOR]

In the era of modern technology, the demand for seamless and fast communication service is gradually increasing. Moreover, as time goes on, more devices are getting connected to the network. In the vast network, for quick responses to these devices, the delay becomes one of the important factors to concern about. Among different types, Queuing delay causes more impact on a network compared to other delays. A survey on delay, different models, effects, and management are illustrated here. Queuing models, like Little Theorem, M/M/1, M/M/m, M/G/1, etc, are discussed in detail. We move on explaining the delay effects on different types of the network such as Wireless network, Mobile communication network, SDN, IoT, etc. This paper also focuses on TCP/IP fields like ECN bits, MPLS, and characteristics of flows. Lastly, the effects of delay on other related networking are discussed. [ABSTRACT FROM AUTHOR]

The increasing growth of online social networks has drawn researchers' attention to link prediction and has been adopted in many fields, including computer sciences, information science, and anthropology. The link prediction in attributed networks is a new challenge in this field, one of the interesting topics in recent years. Nodes are also accompanied in many real-world systems by various attributes or features, known as attributed networks. One of the newest methods of link prediction is embedding methods to generate the feature vector of each node of the graph and find unknown connections. The DeepWalk algorithm is one of the most popular graph embedding methods that capture the network structure using pure random walking. The present paper seeks to present a modified version of deep walk based on pure random walking for solving link prediction in the attributed network, which will be used for both network structure and node attributes, and the new random walk model for link prediction will be introduced by integrating network structure and node attributes, based on the assumption that two nodes on the network will be linked since they are nearby in the network, or connected for the reason of similar attributes. The results indicate that two nodes are more probable to establish a link in the case of possessing more structure and attribute similarity. In order to justify the proposal, the authors carry out many experiments on six real-world attributed networks for comparison with the state-of-the-art network embedding methods. The experimental results from the graphs indicate that our proposed approach is more capable compared to other link prediction approaches and increases the accuracy of prediction. [ABSTRACT FROM AUTHOR]

As a representative model for privacy preserving data publishing, K-anonymity has raised a considerable number of questions for researchers over the past few decades. Among them, how to achieve data release without sacrificing the users' privacy and how to maximize the availability of published data is the ultimate goal of privacy preserving data publishing. In order to enhance the clustering effect and reduce the unnecessary computation, this paper proposes a weighted K-member clustering algorithm. A series of weight indicators are designed to evaluate the outlyingness of records, distance between records, and information loss of the published data. The proposed algorithm can reduce the influence of outliers on the clustering effect and maintain the availability of data to the best possible extent during the clustering process. Experimental analysis suggests that the proposed method generates lower information loss, improves the clustering effect, and is less sensitive to outliers as compared with some existing methods. [ABSTRACT FROM AUTHOR]

With growing heterogeneity and complexity in applications, demand to design an energy-efficient and fast computing system in multi-core architecture has heightened. This paper presents a regression-based dynamic voltage frequency scaling model which studies and utilizes workload characteristics to obtain optimal voltage–frequency (v–f) settings. The proposed framework leverages the workload profile information together with power constraints to compute the best-suited voltage–frequency (v–f) settings to (a) maintain global power budget at chip-level, (b) maximize performance while enforcing power constraints at the per-core level. The presented algorithm works in conjunction with the workload characterizer and senses change in application requirements and apply the knowledge to select the next setting for the core. Our results when compared with two state-of-the-art algorithms MaxBIPS and TPEq achieve the average power reduction of 33% and 25% respectively across 32-core architecture for PARSEC benchmarks. [ABSTRACT FROM AUTHOR]

Considering the complexity of wireless sensor network (WSN) coverage problems, which include many variables and a large continuous search space, a WSN coverage optimization method based on an improved artificial bee colony (ABC) algorithm with teaching strategy is proposed in this paper. ABC, which is good at exploration but poor at exploitation, is improved by introducing a teaching strategy in teaching-learning-based optimization (TLBO) that has a rapid convergence but is easily trapped in a local optima. Thus, the proposed algorithm combines the advantages of ABC strong global search ability and TLBO rapid convergence. In addition, to retain the diversity and eliminate the parameter limit in ABC, a dynamic search update strategy is introduced instead of the scout bee phase of ABC. In addition to preliminary examinations with a number of benchmark functions, the performance of the algorithm is verified by solving a complicated wireless sensor network coverage problem. The simulation results verify that the proposed algorithm achieves better balance between global and local search compared with other state-of-the-art algorithms. [ABSTRACT FROM AUTHOR]

The rapid growth of social networks facilitates the exchange of information whereas malicious behaviors are also steadily increasing in these ecosystems. This results in a challenging situation for individuals to trust other parties. This paper studies the propagation of trust within a chain of trust relations to calculate the trust values of existing users. In this research, an approach for the precise selection of trustworthiness paths as well as the integration of indirect trust values based on the most reliable routes is introduced. The presented approach fuses the ideas from the A* algorithm and multi-criteria decision making approaches using (i.e. TOPSIS method) under fuzzy environments for finding the most reliable path. Moreover, for selecting the most appropriate middle node, a set of criteria such as topological similarity, profile similarity, Dunbar's theorem, local trust, and contextual trust are considered. The evaluation results of the proposed approach demonstrate the propagated trust distance with the different average path lengths while preserving the accuracy of the inferred trust values between each unconnected pair of nodes. The evaluations are performed using the Facebook and Twitter networks having different topologies and the results are compared to the TidalTrust and the MoleTrust algorithms. [ABSTRACT FROM AUTHOR]

Traffic congestions in urban road traffic networks originate from some crowded road segments with crucial locations, and diffuse towards other parts of the urban road network creating further congestions. This behavior of road networks motivates the need to understand the influence of individual road segments on others in terms of congestion. In this paper, we investigate the problems of global influence ranking and local influence ranking of road segments. We propose an algorithm called RoadRank to compute the global influence scores of each road segment from their traffic measures, and rank them based on their overall influence. To identify the locally influential road segments, we also propose an extension called distributed RoadRank, based on road network partitions. We perform extensive experiments on real SCATS datasets of Melbourne. We found that the segments of Batman Avenue, Footscray Road, Punt Road, La Trobe Street, and Victoria Street, are highly influential in the early morning times, which are well known as congestion hotspots for both the network operators and the commuters. Our promising results and detailed insights demonstrate the efficacy of our method. [ABSTRACT FROM AUTHOR]

Community detection plays an important role in the analysis of complex networks. However, overlapping community detection in real networks is still a challenge. To address the problems of pre-input parameters and label redundancy, an improved label propagation algorithm (ILPA) that adopts a method based on the influence factor is proposed in this paper. Theoretical analysis and experimental results on both synthetic and real datasets show that the ILPA detects that the overlapping community has higher accuracy compared to other existing methods. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) has recently received considerable interest due to the development of smart technologies in today's interconnected world. With the rapid advancement in Internet technologies and the proliferation of IoT sensors, myriad systems and applications generate data of a massive volume, variety and velocity which traditional databases and systems are unable to manage effectively. Many organizations need to deal with these massive datasets that encounter different types of data (e.g., IoT streaming data, static data) in different formats (e.g., structured, semi-structured) coming from multiple sources. Several data integration mechanisms have been designed to process mostly static data. Unfortunately, these techniques are not able to deal with and integrate IoT streaming datasets from multiple sources. In this paper, we identify the challenges of IoT Streaming Data Integration (ISDI) and present a formal approach for the real-time integration of such IoT streaming datasets. We address one of the important issues of timing conflict/alignment among streaming data coming from multiple sources. A generic window-based ISDI approach is proposed to deal with IoT data in different formats and algorithms are developed to integrate IoT streaming data from multiple sources. In particular, we extend the basic windowing algorithm for real-time data integration and to deal with the timing alignment issue. We also introduce a de-duplication algorithm to deal with data redundancy and to demonstrate the useful fragments of the integrated data. We conduct several sets of experiments and quantify the performance of our proposed window-based approach. In particular, we compare our local experimental results with a real setup for streaming data, using Apache Spark. The results of the experiments, which are performed on several IoT datasets, show the efficiency of our proposed solution in terms of processing time. The results are also used to provide an integrated data view to the users. [ABSTRACT FROM AUTHOR]

Cloud computing provides utility computing in which clients pay the cost according to their demands and service use. There are some challenges to this technology. One of these issues in data centers is virtual machine (VM) placement so that mapping of these VMs to hosts is executed for a variety of objectives such as load balancing, reducing energy consumption, increasing resource utilization, shortening response time, etc. In this paper, a strategy is presented based on machine learning for VM replacement which aims to balance the load in host machines (HM). In this proposed strategy, the learning agent, in each learning episode by selecting an action from among the permissible actions and executing it on the environment receives a reward according to the desirability of the solution obtained by doing that action in the environment. Receiving a reward from the environment and updating the action value table enable the learner agent to learn in the following episodes that in each environment state, selecting and executing which action is better in the environment and this leads to further enhancement. Our proposed algorithm has, on average, improved the inter-HM load balance in terms of processor, memory, and bandwidth by 25%, 34%, and 32%, respectively, prior to the implementation of the algorithm. Our strategy was compared from diffrent aspects in three scenarios to the MOVMrB strategy. Finally, it was concluded that our proposed algorithm can be more effective in load balancing by having much less runtime and turning off more HMs. [ABSTRACT FROM AUTHOR]

In a large-scale data center, it is vital to precisely recognize the termination statuses of applications at an early stage. In recent years, many machine learning techniques have been applied to this issue, which is beneficial for optimizing the scheduling policy and improving the efficiency of resource utilization. However, if the application's dynamic information is insufficient at the early stage, the generalization performance of the machine learning model will be lessened, and the prediction accuracy could be low. To overcome this problem, a novel failure prediction method that is based on the association relationships between similar jobs is proposed in this paper to jointly predict task's termination statuses at an earlier stage. The similar jobs whose tasks have similar changing modes of consumed resources, an inherent structural correlation may exist, and the correlation information is significant for improving the prediction model's generalization performance. First, a job clustering algorithm is proposed for identifying the jobs with higher similarity from jobs that have various numbers of tasks. Second, based on the job clustering results, the robust multi-task learning algorithm is introduced to effectively utilize the domain information among jobs (i.e. interactional relationship among jobs on the termination statuses of task). Experiments are conducted on a Google cluster workload traces dataset. The results show that the proposed method can realize higher prediction accuracy, lower misjudgment rate, and higher predictive stability than several state-of-the-art methods at 1/3 the running time of the tasks. [ABSTRACT FROM AUTHOR]

For automating the deployment of applications in cloud environments, a variety of technologies have been developed in recent years. These technologies enable to specify the desired deployment in the form of deployment models that can be automatically processed by a provisioning engine. However, the deployment across several clouds increases the complexity of the provisioning. Using one deployment model with a single provisioning engine, which orchestrates the deployment across the clouds, forces the providers to expose low-level APIs to ensure the accessibility from outside. In this paper, we present an extended version of the split and match method to facilitate the division of deployment models to multiple models which can be deployed by each provider separately. The goal of this approach is to reduce the information and APIs which have to be exposed to the outside. We present a formalization and algorithms to automate the method. Moreover, we validate the practical feasibility by a prototype based on the TOSCA standard and the OpenTOSCA ecosystem. [ABSTRACT FROM AUTHOR]

Most recommender systems suggest items that are popular among all users and similar to items a user usually consumes. As a result, the user receives recommendations that she/he is already familiar with or would find anyway, leading to low satisfaction. To overcome this problem, a recommender system should suggest novel, relevant and unexpected i.e., serendipitous items. In this paper, we propose a serendipity-oriented, reranking algorithm called a serendipity-oriented greedy (SOG) algorithm, which improves serendipity of recommendations through feature diversification and helps overcome the overspecialization problem. To evaluate our algorithm, we employed the only publicly available dataset containing user feedback regarding serendipity. We compared our SOG algorithm with topic diversification, popularity baseline, singular value decomposition, serendipitous personalized ranking and Zheng's algorithms relying on the above dataset. SOG outperforms other algorithms in terms of serendipity and diversity. It also outperforms serendipity-oriented algorithms in terms of accuracy, but underperforms accuracy-oriented algorithms in terms of accuracy. We found that the increase of diversity can hurt accuracy and harm or improve serendipity depending on the size of diversity increase. [ABSTRACT FROM AUTHOR]

Spatial indexes are important in spatial databases for efficient execution of queries involving spatial constraints. The xBR + -tree is a balanced, disk-resident, Quadtree-based index for point data, which is very efficient for processing spatial queries. Bulk-loading refers to the process of creating an index from scratch as a whole, when the dataset to be indexed is available beforehand, instead of creating (loading) the index gradually, when the dataset items are available one-by-one. Bulk insertion refers to the process of updating an existing index by inserting a large batch of new data, treating the items of this batch as a whole and not by inserting these items one-by-one. In this paper, we modify previous bulk-loading and bulk-insertion algorithms for xBR + -trees to achieve higher performance by taking advantage of the special features of Solid State Drives (SSDs). SSDs have attracted database developers, mainly due to their higher read performance (thanks to their internal parallelism) than Hard Disk Drives. Using real and artificial datasets of various cardinalities, we experimentally compare the modified algorithms against their predecessors and show that the modified algorithms are clear winners regarding performance. [ABSTRACT FROM AUTHOR]

Thinning algorithms have been widely applied in many applications such as computer graphics, computer vision and medical imaging. The performances of thinning algorithms are hampered by the increase of image's size and image's content. By employing graphics processing units (GPUs), the calculation of thinning algorithms can be accelerated. In this paper, we present GPU parallel versions of well-known binary thinning algorithms namely the ZS algorithm, the GH algorithm, the AW algorithm and the hybrid algorithm belonging to different classes for an efficient implementation using CUDA as a parallel programming model. These thinning algorithms are then evaluated and compared according to quality and runtime measures. The obtained results show that the GPU multithreaded implementations of the selected algorithms present high computational speed outperforming the central processing unit sequential implementations achieving the best average speedup when executing the AW algorithm. [ABSTRACT FROM AUTHOR]