Academic & Scientific Articles
Permanent URI for this communityhttp://dl.cerist.dz/handle/CERIST/3
Browse
39 results
Search Results
Item New GPU-based Swarm Intelligence Approach For Reducing Big Association Rules Space(CERIST, 2017-06-14) Djenouri, Youcef; Bendjoudi, Ahcène; Djenouri, Djamel; Belhadi, Asma; Nouali-Taboudjemat, NadiaThis paper deals with exploration and mining of association rules in big data, with the big challenge of increasing computation time. We propose a new approach based on meta-rules discovery that gives to the user the summary of the rules’ space through a meta-rules representation. This allows the user to decide about the rules to take and prune. We also adapt a pruning strategy of our previous work to keep only the representatives rules. As the meta-rules space is much larger than the rules space, two approaches are proposed for efficient exploitation. The first one uses a bees swarm optimization method in the meta-rules discovery process, which is extended using GPU-based parallel programming to form the second one. The sequential version has been first tested using medium rules set, and the results show clear improvement in terms of the number of returned meta-rules. The two versions have then been compared on large scale rules sets, and the results illustrate the acceleration on the summarization process by the parallel approach without reducing the quality of resulted meta-rules. Further experiments on Webdocs big data instances reveal that the proposed method of pruning rules by summarizing meta-rules considerably reduces the association rules space compared to state-of-the-art association rules mining-based approaches.Item GPU-based Bio-inspired Model for Solving Association Rules Mining Problem(CERIST, 2017-03-06) Djenouri, Youcef; Bendjoudi, Ahcène; Djenouri, Djamel; Commuzi, Marcoproblem with the purpose of extracting the correlations between items in sizeable data instances. According to the state of the art, the bio-inspired approaches proved their usefulness by finding high number of satisfied rules in a reasonable time when dealing with medium size instances. These approaches are unsuitable for large databases and especially for those existing on the web such as the Webdocs instance. Recently, the Graphics Processor Units (GPU) is considered as one of the most used parallel hardware to solve large scientific complex problems. In this paper, we propose a new GPU-based model of the bio-inspired approaches for solving association rules mining problem. Our model benefits from the massively GPU threaded by evaluating multiple rules in parallel on GPU. To validate the proposed model, the most used bio-inspired approaches (GA, PSO, and BSO) have been executed on GPU to solve well-known large ARM instances. Real experiments have been carried out on an Intel Xeon 64 bit quad-core processor E5520 coupled to an Nvidia Tesla C2075 GPU device. The results show that the genetic algorithm outperforms PSO and BSO. Moreover, it outperforms the state-of-the-art GPU-based ARM approaches when dealing with the challenging Webdocs instance.Item Impact of Genetic Algorithms Operators on Association Rules Extraction(CERIST, 2016-10-02) Hamdad, Leila; Ournani, Zakaria; Benatchba, Karima; Bendjoudi, AhcèneIn this paper, we study the impact of GAs’ components such as encoding, different crossover, mutation and replacement strategies on the number of extracted association rules and their quality. Moreover, we propose a strategy to manage the population. The later is organized in classes where each one encloses same size rules. Each class can be seen as a population on which a GA is applied. All tests are conducted on two types of benchmarks : synthetic and real ones of different sizes.Item Impact of Genetic Algorithms Operators on Association Rules Extraction(2016-11-11) Hamdad, Leila; Ournani, Zakaria; Benatchba, Karima; Bendjoudi, AhcèneIn this paper we study the impact of GAs’ components such as encoding, different crossover, mutation and replacement strategies on the number of extracted association rules and their quality. Moreover, we propose a strategy to manage the population. The later is organized in classes where each one encloses same size rules. Each class can be seen as a population on which a GA is applied. All tests are conducted on two types of benchmarks : synthetic and real ones of different sizes.Item Efficient parallel B&B method for the Blocking Job Shop Scheduling Problem(2016-07) Dabah, Adel; Bendjoudi, Ahcène; Ait Zai, AbdelhakimThe blocking job shop scheduling problem (BJSS) is a version of the classical job shop scheduling with no intermediate buffer between machines. It is known to be NP-hard in the strong sense. The major drawbacks of the previous works are the huge time needed to explore the search space and the low ratio of feasible to explored solutions when applying metaheuristics, leading to a poor quality of final solutions. In order to accelerate the exploration of the search space and overcome the drawback of metaheuristics, we propose in this paper a parallel Branch and Bound (B&B) algorithm to act both as an exact or approximate methods. The proposed parallel B&B approach is based on the master-worker paradigm, exploiting the computing power offered by cluster architectures. The master performs a breadth-first exploration to ensure high availability of tasks (sub-problems) while the workers perform a depth-first exploration in order to browse a large number of feasible solutions, therefore, obtaining a faster improvement of the solution quality. The proposed approach has been executed on a cluster based architecture using 32 nodes with 16 CPU cores each. The obtained results show a good speedup of the execution time with 80% efficiency. We have been able to solve optimally ten BJSS benchmark instances never solved before. Moreover, our proposed approach improved the best solutions for more than 22 benchmark instances.Item Multi and Many-core Parallel B&B approaches for the Blocking Job Shop Scheduling Problem(2016-07) Dabah, Adel; Bendjoudi, Ahcène; Ait Zai, Abdelhakim; El Baz, Didier; Nouali-Taboudjemat, NadiaIn this paper, we propose three approaches to accelerate the B&B execution time using Multi and Many-core systems to solve the NP-hard Blocking Job Shop Scheduling problem (BJSS). The first approach is based on Master/Worker paradigm where the workers independently explore the branches sent by the master. The second approach is a node-based parallelization that does not change the design of the B&B algorithm, except that the bounding process is faster since it is calculated in parallel using several threads organized in one GPU block. The third approach is a Multi-Core CPU/GPU hybridization that benefits from the power of both the CPU-cores and the GPU at the same time. This hybridization is based on concurrent kernels execution provided by Nvidia Multi process Service (MPS) i.e. each host process (Master or Worker) launches his own kernel to accelerate the bounding process on GPU. The obtained results using Taillard instances confirm the efficiency of our proposals. The first two approaches are respectively three and eighteen times faster compared to the sequential version. The results of the hybrid approach show a relative speedup over ninety times as compared to the sequential approach and therefore prove the advantage of using both the CPU-cores and the GPU at the same time.Item Parallel BSO Algorithm for Association Rules Mining Using Master/Worker Paradigm(2015-09-06) Djenouri, Youcef; Bendjoudi, Ahcène; Djenouri, Djamel; Habbas, ZinebThe extraction of association rules from large transactional databases is considered in the paper using cluster architecture parallel computing. Motivated by both the successful sequential BSO-ARM algorithm, and the strong matching between this algorithm and the structure of the cluster architectures, we present in this paper a new parallel ARM algorithm that we call MW-BSO-ARM for Master/Workers version of BSO-ARM. The goal is to deal with large databases by minimizing the communication and synchronization costs, which represent the main challenges that faces any cluster architecture. The experimental results are very promising and show clear improvement that reaches 300% for large instances. For examples, in big transactional database such as WebDocs, the proposed approach generates 107 satisfied rules in only 22 minutes, while a previous GPU-based approach cannot generate more than 103 satisfied rules into 10 hours. The results also reveal that MWBSO-ARM outperforms the PGARM cluster-based approach in terms of computation time.Item Reducing thread divergence in GPU-based bees swarm optimization applied to association rule mining(John Wiley & Sons, Ltd., 2016) Bendjoudi, Ahcène; Djenouri, Youcef; Habbas, Zineb; Mehdi, Malika; Djenouri, DjamelThe association rules mining (ARM) problem is one of the most important problems in the area of data mining. It aims at finding all relevant association rules from transactional databases. It is CPU time intensive and requires a huge computing power when dealing with large transactional databases. To deal with this issue, Graphics Processing Units (GPUs) are a powerful tool to speed up the search process. However, their performance is closely subject to thread/branch divergence resulting from the single instruction multiple data parallel model of GPUs. In this paper, we propose three approaches based on database reorganization, aiming to reduce thread divergence in GPU-based bees swarm optimization metaheuristic for ARM, respectively, named block-based reordering, transactions-based reordering, and transactions-based reordering with median value. Theoretical and experimental studies have been carried out using well-known large ARM instances. The experiments have been performed on an Intel Xeon 64 bit quad-core processor E5520 coupled to Nvidia Tesla C2075 448 cores. The results show that the proposed approaches minimize considerably the number of thread divergence and improve the overall execution time. Indeed, the number of thread divergence occurrences has been reduced by up to eight times making the execution much faster.Item GPU parallel B&B for the Blocking job shop scheduling problem.(CERIST, 2016-02-22) Dabah, Adel; Bendjoudi, Ahcène; Ait Zai, Abdelhakim; El Baz, DidierBranch and bound algorithms (B&B) are well known techniques for solving optimally combinatorial optimization problems. Nevertheless, these algorithms remain inefficient when dealing with large instances. This paper deals with the blocking job shop scheduling (BJSS), which is a version of classical job shop scheduling with no intermediate buffer between machines. This problem is an NP-hard problem and its exact resolution using the sequential approach is impractical. We propose in this paper a GPU-based parallelization in which a two level scheme is used. The first level is a node-based parallelization in which the bounding process is faster because it is calculated in parallel using several threads organized in one GPU block. To fully occupy the GPU, we propose a second level of parallelization which is a generalization of the first level. Therefore, for each iteration several search tree nodes are evaluated on the GPU using several thread-blocks. The obtained results, using the well-known Taillard instances, confirm the efficiency of the proposed approach. Also, the results show that our approach is 65 times faster than an optimized sequential B&B version.Item GPU-based two level parallel B&B for the Blocking job shop scheduling problem.(IEEE, 2016-05-23) Adel, Dabah; Bendjoudi, Ahcène; El Baz, Didier; Abdelhakim, Ait ZaiBranch and bound algorithms (B&B) are well known techniques for solving optimally combinatorial optimization problems. Nevertheless, these algorithms remain inefficient when dealing with large instances. This paper deals with the blocking job shop scheduling (BJSS), which is a version of classical job shop scheduling with no intermediate buffer between machines. This problem is an NP-hard problem and its exact resolution using the sequential approach is impractical. We propose in this paper a GPU-based parallelization in which a two level scheme is used. The first level is a node-based parallelization in which the bounding process is faster because it is calculated in parallel using several threads organized in one GPU block. To fully occupy the GPU, we propose a second level of parallelization which is a generalization of the first level. Therefore, for each iteration several search tree nodes are evaluated on the GPU using several thread-blocks. The obtained results, using the well-known Taillard instances, confirm the efficiency of the proposed approach. Also, the results show that our approach is 65 times faster than an optimized sequential B&B version.