期刊名称：International Journal of Advances in Soft Computing and Its Applications
印刷版ISSN：2074-8523
出版年度：2020
卷号：12
期号：1
页码：15-34
出版社：International Center for Scientific Research and Studies
摘要：Classification remains as a most significant area in data mining.Probabilistic Neural Network (PNN) is repeatedly used for classificationproblems. The main aims of this paper are to fine-tune the neural networksweights to increase the classification accuracy and to achieve speedconvergence. To achieve this aim, created a hybrid model that investigate theFlower Pollination Algorithm (FPA) with PNN, where the PNN generated theinitial solutions randomly and then the FPA is used to adjust the weight of thePNN, and further improvements were conducted using the FPA, whichoptimized the PNN weights. Experimental results using 11 benchmark datasetsshowed that the proposed FPA with PNN perform better than the original PNNon all data sets. After compared with another algorithm in the past work fromthe literature, the FPA can get improved results in regarding of theclassification accuracy.
