The emergence of GPU-CPU heterogeneous architecture has led to a significant paradigm shift in parallel programming. How to effectively implement parallelgenetic Algorithm (GA) in these environments has become one of...
详细信息
The emergence of GPU-CPU heterogeneous architecture has led to a significant paradigm shift in parallel programming. How to effectively implement parallelgenetic Algorithm (GA) in these environments has become one of the current hot issues. GA's calculation and operators are closely related to specific problems, thereby significantly affecting the acceleration method of GA algorithms. The Generalized Assignment Problem (GAP) is a classic NP-hard combinatorial optimization problem. The more widely used geneticalgorithms to solve the GAP in the CPU are difficult to be parallelized in a GPU environment due to severe data dependencies. To address this problem, two algorithms suitable for the implementation on the GPU are proposed, namely RPE algorithm and NNE algorithm, which obtain significant performance speedup by alleviating data dependencies and mutually exclusive synchronization overheads. At the same time, considering the new GPU architecture features and programming models, three different granular implementations of parallel genetic algorithms to solve the GAP are proposed, namely GPGA(thread), GPGA(warpsp) and GPGA(cgroup), by utilizing the warp-specialization technology and the cooperative group mechanism. GPGA series algorithms obtain better solution quality and very significant performance improvements compared with Serial GA, GTS (the GPU-CPU hybrid implementation of Scatter Search with Tabu lists) and Lagrange Relaxation algorithm on a CPU by solving 16 typical large-scale GAP instances.
The need to improve the scalability of geneticalgorithms (GAs) has motivated the research on parallel genetic algorithms (PGAs), and different technologies and approaches have been used. Hadoop MapReduce represents o...
详细信息
The need to improve the scalability of geneticalgorithms (GAs) has motivated the research on parallel genetic algorithms (PGAs), and different technologies and approaches have been used. Hadoop MapReduce represents one of the most mature technologies to develop parallelalgorithms. Based on the fact that parallelalgorithms introduce communication overhead, the aim of the present work is to understand if, and possibly when, the parallel GAs solutions using Hadoop MapReduce show better performance than sequential versions in terms of execution time. Moreover, we are interested in understanding which PGA model can be most effective among the global, grid, and island models. We empirically assessed the performance of these three parallel models with respect to a sequential GA on a software engineering problem, evaluating the execution time and the achieved speedup. We also analysed the behaviour of the parallel models in relation to the overhead produced by the use of Hadoop MapReduce and the GAs' computational effort, which gives a more machine-independent measure of these algorithms. We exploited three problem instances to differentiate the computation load and three cluster configurations based on 2, 4, and 8 parallel nodes. Moreover, we estimated the costs of the execution of the experimentation on a potential cloud infrastructure, based on the pricing of the major commercial cloud providers. The empirical study revealed that the use of PGA based on the island model outperforms the other parallel models and the sequential GA for all the considered instances and clusters. Using 2, 4, and 8 nodes, the island model achieves an average speedup over the three datasets of 1.8, 3.4, and 7.0 times, respectively. Hadoop MapReduce has a set of different constraints that need to be considered during the design and the implementation of parallelalgorithms. The overhead of data store (i.e., HDFS) accesses, communication, and latency requires solutions that reduce data store o
Making geneticalgorithms (GAs) distributed in an on demand fashion involves different phases from resources allocation to actual deployment and execution. We propose a cloud architecture with a conceptual workflow ab...
详细信息
ISBN:
(纸本)9781450343237
Making geneticalgorithms (GAs) distributed in an on demand fashion involves different phases from resources allocation to actual deployment and execution. We propose a cloud architecture with a conceptual workflow able to cover each GAs distribution phase.
elephant56 is an open source framework for the development and execution of single and parallel genetic algorithms (GAs). It provides high level functionalities that can be reused by developers, who no longer need to ...
详细信息
ISBN:
(纸本)9781450343237
elephant56 is an open source framework for the development and execution of single and parallel genetic algorithms (GAs). It provides high level functionalities that can be reused by developers, who no longer need to worry about complex internal structures. In particular, it offers the possibility of distributing the GAs computation over a Hadoop MapReduce cluster of multiple computers. In this paper we describe the design and the implementation details of the framework that supports three different models for parallel GAs, namely the global model, the grid model and the island model. Moreover, we provide a complete example of use.
geneticalgorithms (GAs) can be the tool of choice especially for optimizing combinatorial and complex problems in transport and infrastructure systems such as traffic signal control, pavement rehabilitation and desig...
详细信息
geneticalgorithms (GAs) can be the tool of choice especially for optimizing combinatorial and complex problems in transport and infrastructure systems such as traffic signal control, pavement rehabilitation and design, and transit service scheduling. This paper presents an overview of different techniques to improve performance of GAs, with particular emphasis on parallel GAs (PGAs). Results are presented from applications of a simple GA (SGA) and a migration PGAs on a traffic control problem, a benchmark GA-difficult, and benchmark GA-easy problem. For all problems, savings in computation resources were realized when PGA was used. Advantages of PGAs are more pronounced for complex and difficult (deceptive) problems. On a difficult problem tested in this research, a PGA with four subpopulations was 7times more efficient than a serial one, and a PGA with eight subpopulations was more than 18times more efficient. With smaller and less complex problems, the impact of parallelism is less dramatic when the computation resources are limited. Use of parallel GAs does not reduce the importance of seeking efficient problem-specific operators and parameter values, but does magnify the effectiveness of such choices and increase the range of options available. The advantages PGAs offer mean more efficient and faster optimization for many applications in civil infrastructure design, operating management, and maintenance projects.
This paper describes a framework for developing parallel genetic algorithms (GAs) on the Hadoop platform, following the paradigm of MapReduce. The framework allows developers to focus on the aspects of GA that are spe...
详细信息
ISBN:
(纸本)9781450331968
This paper describes a framework for developing parallel genetic algorithms (GAs) on the Hadoop platform, following the paradigm of MapReduce. The framework allows developers to focus on the aspects of GA that are specific to the problem to be addressed. Using the framework a GA application has been devised to address the Feature Subset Selection problem. A preliminary performance analysis showed promising results.
This paper describes an efficient solution to parallelize softwareprogram instructions, regardless of the programming language in which theyare written. We solve the problem of the optimal distribution of a set ofinst...
详细信息
This paper describes an efficient solution to parallelize softwareprogram instructions, regardless of the programming language in which theyare written. We solve the problem of the optimal distribution of a set ofinstructions on available processors. We propose a genetic algorithm to parallelize computations, using evolution to search the solution space. The stagesof our proposed genetic algorithm are: The choice of the initial populationand its representation in chromosomes, the crossover, and the mutation operations customized to the problem being dealt with. In this paper, geneticalgorithms are applied to the entire search space of the parallelization ofthe program instructions problem. This problem is NP-complete, so thereare no polynomial algorithms that can scan the solution space and solve theproblem. The genetic algorithm-based method is general and it is simple andefficient to implement because it can be scaled to a larger or smaller number ofinstructions that must be parallelized. The parallelization technique proposedin this paper was developed in the C# programming language, and our resultsconfirm the effectiveness of our parallelization method. Experimental resultsobtained and presented for different working scenarios confirm the theoreticalresults, and they provide insight on how to improve the exploration of a searchspace that is too large to be searched exhaustively.
Ensuring the security of sensitive or private information is crucial to prevent malicious tampering, especially in multimedia applications like intelligent transport systems (ITS), which are vital components of a smar...
详细信息
Ensuring the security of sensitive or private information is crucial to prevent malicious tampering, especially in multimedia applications like intelligent transport systems (ITS), which are vital components of a smart city. These systems can be vulnerable to traffic management and rerouting techniques that manipulate the images captured by roadside units. To address this challenge, this paper introduces advanced image encryption algorithms designed specifically for securing image manipulation and transmission in roadside ITS units. Initially, a sequential version of the algorithm is proposed, demonstrating a high level of confusion achieved through the chosen coding method (chromosomal representation). This sequential approach results in maximum interference between the original image and its encrypted counterpart, with an entropy level of 7.95, nearing the optimal value of 8. To improve computational efficiency, three additional algorithms are presented, utilizing parallelization based on the islanding model, both with and without migrations. The algorithms are designed to enhance security by increasing confusion and incorporating genetic diffusion. The performance and security of these algorithms are evaluated using established methods such as information entropy, differential attack analysis, and key space analysis. Our algorithms have also shown a strong ability to maintain performance and robustness even in the presence of noise. Furthermore, they exhibit superior resistance to attacks compared to recent competitive approaches. In summary, the proposed algorithms offer robust protection against image manipulation and unauthorized access in roadside ITS units, thereby contributing to the overall security and reliability of smart city infrastructure.
The Vehicle Routing Problem (VRP) is fundamental to logistics operations. Finding optimal solutions for VRPs related to large, real-world operations is computationally expensive. geneticalgorithms (GA) have been used...
详细信息
ISBN:
(数字)9781665497862
ISBN:
(纸本)9781665497862
The Vehicle Routing Problem (VRP) is fundamental to logistics operations. Finding optimal solutions for VRPs related to large, real-world operations is computationally expensive. geneticalgorithms (GA) have been used to find good solutions for different types of VRPs but are slow to converge. This work utilizes high-performance computing (HPC) platforms to design a parallel GA (PGA) algorithm for solving large-scale VRP problems. The algorithm is implemented on an eight-GPU NVIDIA DGX-1 server. Maximum parallelism is achieved by mapping all algorithm arrays into block threads to achieve high throughput and reduced latency for full GPU utilization. Tests with VRP benchmark problems of up to 20,000 nodes compare the algorithm performance (speed) with different GPU counts and a multi-CPU implementation. The developed algorithm provides the following improvements over CPU or single-GPUbased algorithms: (i) larger problem sizes up to 20,000 nodes are handled, (ii) execution time is reduced over the CPU by a factor of 1,700, and iii) for the range tested, the performance increases monotonically with the number of GPUs.
Pattern recognition has been evolving to include problems posed by new sceneries containing a high number of pattern components. Processing this volume of information allows a more exact classification in wider types ...
详细信息
ISBN:
(纸本)9783031477645;9783031477652
Pattern recognition has been evolving to include problems posed by new sceneries containing a high number of pattern components. Processing this volume of information allows a more exact classification in wider types of applications;however, some of the difficulties of this scheme is the maintenance of numerical precision and mainly the reduction of the execution time. During the last 15 years, several Machine Learning solutions have been implemented to reduce the number of pattern components to be analyzed, such as artificial neural networks. Deep learning is an appropriate tool to accomplish this task. In this paper, a convolutional neural network is implemented for recognition and classification of human activity signals and digital images. It is achieved by automatically adjusting the parameters of the neural network through geneticalgorithms using a multiprocessor and GPU platform. The results obtained show the reduction of computational costs and the possibility of better understanding of the solutions provided by Deep Learning.
暂无评论