基于模糊聚类的神经网络在数据挖掘分类中的应用研究

常见的决策树分类算法、贝叶斯分类算法、神经网络分类算法为数据挖据分类算法研究提供了重要基础。但面对海量数据时,在时间效率、鲁棒性和精确性上都显示出了不足。为此,本文将模糊聚类的思想引入到神经网络分类算法中,首先通过模糊聚类子模型,将样本数据聚为几个数据子集,然后再采用不同的神经网络对各个数据子集同时进行训练学习。由于经过了模糊聚类子模型的预处理,每个神经网络训练学习样本的复杂性大大减少,使神经网络的学习效率大大提高。最后通过UCI下的实际数据库,对提出的分类算法进行了检验,结果显示了基于模糊聚类的神经网络在数据挖掘分类中应用的有效性。

Application of Neural Network Based on Fuzzy Clustering in the Classification Algorithm of Data Mining

The common classification algorithms, such as Decision Tree, Bayesian classification and Artificial Neural Network, provide an important foundation for the classification algorithm of data mining. But due to huge amounts of data, these algorithms face some challenges in time efficiency, robustness and accuracy. This paper proposes an artificial neural network model based on fuzzy clustering. First, it clusters the training data into different sub-data using the Fuzzy Clustering Model. Subsequently, it uses various artificial neural networks to train the sub-data. After that, the number and complexity of training data is reduced and the efficiency of the artificial neural network is enhanced greatly. In the end, the UCI's databases are used to prove the usefulness of the new classification algorithm. The results show the validity of application of the Neural Network based on fuzzy clustering in the classification of data mining.