面向大数据的增量式RBF学习算法

RBF Incremental Learning Algorithm for Big Data

"大数据"背景下,如何处理庞大数据成为众多企业关注的热点。文章提出了一种新的在线处理大数据的方法,利用数学中的分块矩阵定理对径向基函数进行处理,从在线处理大数据的角度思考,利用增量学习算法原理推导出径向基函数(RBF)增量学习算法模型,为大数据的增量算法提供一种新思路,并利用实际算例加以检验。实验表明,相对于传统的一次性建模的方法,所提出的增量式RBF算法能在保证不影响建模精度的前提下明显地缩短处理大数据的时间。

Under the background of Big Data, how to deal with the huge data has became the focus of attention of many enterprises. This paper presents a new method of online data processing, which uses the partition matrix theorem in mathematics to process radial basis function(RBF). Considering from the perspective of processing big data online, the paper uses the principle of incremental learning algorithm to deduce the incremental RBF learning model, providing a new idea for incremental algorithm of big data. Finally, a practical example is used to make verification. Experimental results show that compared with the traditional one-off modeling method, the proposed incremental RBF algorithm can significantly shorten the processing time of big data without affecting the modeling accuracy.