基于MapReduce改进离散型萤火虫算法和多重分形的属性约简方法

Attribute Reduction Method Based on MapReduce-Based Improved Discrete Glowworm Swarm Algorithm and Multi-fractal Dimension

针对大数据环境下属性约简问题,提出基于MapReduce改进离散型萤火虫算法(IDGSO)和多重分形(MFD)的属性约简方法.首先,通过对萤火虫个体的移动方式进行离散化改进,引入迁徙策略和高斯变异策略,避免陷入局部最优,并提出改进离散型萤火虫算法.然后,将IDGSO结合MFD应用于属性约简中.最后,针对大数据环境下属性约简问题,采用MapReduce编程模式,实现对IDGSO和MFD的并行化.在UCI数据集和实际气象数据集上的实验表明,文中算法约简性能较优,运行效率较快,具有较好的有效性和可行性.

To solve the problem of attribute reduction in a big data environment, an attribute reduction method based on MapReduce-based improved discrete glowworm swarm algorithm（IDGSO） and muhi-fractal dimension （MFD） is proposed. Firstly, the moving way of glowworm individuals is discretized to avoid the algorithm falling into local optimum, and the migration strategy and Gaussian mutation strategy are introduced. An improved discrete glowworm swarm algorithm is proposed. Secondly, the improved discrete glowworm algorithm combined with multi-fractal dimension is applied to attribute reduction. Finally, to solve the problem mentioned above, the MapReduce programming model is adopted to realize the parallelization of IDGSO and MFD. Experiments on UCI datasets and the real meteorological datasets show that the proposed method produces high efficiency, effectiveness and feasibility of reduction.