ConvGeN: A convex space learning approach for deep-generative oversampling and imbalanced classification of small tabular datasets

Kristian Schultz, Saptarshi Bej, Waldemar Hahn, Markus Wolfien, Prashant Srivastava, Olaf Wolkenhauer

Abstract: Highlights•Deep adversarial learning isn’t apt for oversampling on small tabular imbalanced data.•ConvGeN learns appropriate convex coefficients from each minority data neighborhood.•ConvGeN is the first deep learning architecture for interpolation-based oversampling.•ConvGeN can by itself learn which parts of the convex data space could be important.•ConvGeN can be useful especially when there is a small amount of training data.