Cost-Sensitive Variational Autoencoding Classifier for Imbalanced Data Classification

Classification is among the core tasks in machine learning. Existing classification algorithms are typically based on assumption of at least roughly balanced data classes. When performing involving imbalanced data, such classifiers ignore minority consideration overall accuracy. The performance traditional distribution insufficient because minority-class samples often more important than others, as positive samples, disease diagnosis. In this study, we propose a cost-sensitive variational autoencoding classifier that combines data-level and algorithm-level methods to solve problem classification. Cost-sensitive factors introduced assign high cost misclassification which biases toward data. We also designed costs closely related by embedding domain knowledge. Experimental results show proposed method performed bulk amorphous materials well.