Quantum-Behaved Particle Swarm Optimization Using MapReduce
Quantum-behaved particle swarm optimization (short in QPSO) is an improved version of particle swarm particle (short in PSO), and the performance is superior. But for now, it may not always satisfy the situations. Nowadays, problems become larger and more
- PDF / 111,854 Bytes
- 6 Pages / 439.37 x 666.142 pts Page_size
- 62 Downloads / 261 Views
Abstract. Quantum-behaved particle swarm optimization (short in QPSO) is an improved version of particle swarm particle (short in PSO), and the performance is superior. But for now, it may not always satisfy the situations. Nowadays, problems become larger and more complex, most serial optimization algorithms cannot deal with the problem or need plenty of computing cost. In this paper, we implement QPSO on MapReduce model, propose MapReduce quantum-behaved particle swarm optimization (short in MRQPSO), and realize QPSO parallel and distributed, which the MapReduce model is a parallel computing programming model. In the experiments, the test results show that MRQPSO is more advanced both on performance of solution and time than QPSO. Keywords: Quantum-behaved particle swarm optimization · MapReduce · Distributed evolutionary computation · Cloud computing
1
Introduction
In recent years, many intelligent algorithms are facing with a serious difficulty the more and more large-scale data. Such as web content and bioinformatics data, the simple serial algorithms may be confused when processing these tough problems, needless to say some deceptive ones. In order to deal with hard optimization problems in real-word applications, distributed evolutionary algorithms (short in dEAs) have been blossomed rapidly. In the paper, it provides a comprehensive survey of the EAs and models and discuss the parallel and distributed genetic algorithms in different physical platforms. The particle swarm optimization is an outstanding one of genetic algorithm [1–3]. This algorithm is proposed by Kennedy and Eberhart in 1995 in [4]. Depending on rapid convergence as well as good solution performance, the PSO has been attained increasing attention. However, the premature phenomenon as a drawback may influence the performance of solution. Focus on this shortcoming, Sun proposed the quantum-behaved particle swarm optimization in 2004 in [5] and presented a comprehensive analysis in [6]. Due to the quantum mechanics, c Springer Nature Singapore Pte Ltd. 2016 M. Gong et al. (Eds.): BIC-TA 2016, Part II, CCIS 682, pp. 173–178, 2016. DOI: 10.1007/978-981-10-3614-9 22
174
Y. Li et al.
the QPSO avoids the particle fall in the local optimum greatly. Unfortunately, when the algorithm faces large-scale and complex problem, the increasing computational cost and the still existed premature phenomenon urge the original algorithm to be parallel. In [7], MapReduce, as a tool to be adopted to implement dEAs, is proposed by Google in 2004. To respond the requirement of parallelization and distribution, this physical platform is very convenient to deploy an algorithm to update to be parallel. The programmers only need to consider the map function and reduce function, and the other details are provided by the model itself. Because of the convenience, the scholars can focus on the algorithms and problems, and appear many genetic algorithms through MapReduce to realize distributed, including the PSO [8]. In order to following this trend and enhancing the capabi
Data Loading...
