PROJECT TITLE :

Semi-Supervised Deep Fuzzy C-Mean Clustering for Imbalanced Multi-Class Classification

ABSTRACT:

Semi-supervised learning has been effectively linked in machine learning study topics like data mining and dynamic data analysis. One of the most difficult difficulties for classification is unbalanced class learning. Several scholars have focused their efforts in recent years on data categorization of multi-class imbalanced datasets. For imbalanced multi-class classification, we presented semi-supervised deep Fuzzy C-mean clustering in this study (DFCM-MC). The term "deep" is used in our paper to describe how the decomposition approach is utilized to breakdown semi-supervised data into supervised (labeled) and unsupervised (unlabeled) data. We employed both unlabeled and labeled data to extract discriminative information, which is useful for classification, when training the model. Second, it divides supervised and unsupervised data into multiple intra-clusters to address the problem of multi-class imbalance data, which favors intra-cluster classes and intra-cluster features. We present the DFCM-MC technique, which associates the highest similarity of features between multi-intra clusters and uses multi-intra clusters to extract new features to control redundancy for multi-class imbalance classification. We also use the re-sampling strategy to handle the imbalance data for classification, which improves the DFCM-classification MC's performance. We compare the performance of our proposed technique with four state-of-the-art learning algorithms for multi-class imbalance data using three performance indicators on 18 benchmark multi-class imbalanced datasets (mean of accuracy, mean of f-measure, and mean of area under the curve). The results of the experiments show that our suggested method outperforms others because of its ability to recognize and condense fundamental information from unsupervised data.