A Hybrid PCA-Fuzzy-ELM to Predict QSARs for the Inhibition of Dihydrofolate Reductase by Pyrimidines

Fully connected fuzzy inference systems (F-CONFIS) proposed recently by Wang et.al. when combined with Extreme Learning Machines (ELMs) outperforms other classifiers. We propose a hybrid PCA-Fuzzy-ELM method in this paper. In this hybrid method, firstly the dimension of the dataset is reduced using PCA, secondly a modified fully connected fuzzy inference system (MF-CONFIS) is designed and finally ELM is used to train the MF-CONFIS. The classification results based on benchmarking datasets shows the merits of the proposed hybrid PCA-Fuzzy-ELM. Finally, the proposed hybrid PCA-Fuzzy-ELM is used to Predict QSARs for the Inhibition of Dihydrofolate Reductase by Pyrimidines.