Transfer inhibitory potency prediction to binary classification: A model only needs a small training set

Haowen Dou, Jie Tan, Huiling Wei, Fei Wang, Jinzhu Yang, Xiaoguang Ma, Jiaqi Wang, Teng Zhou

Abstract: Highlights•An inhibitory potency prediction model only needs to be trained on a small dataset.•Transfer the inhibitory potency prediction task to a simpler binary classification task by designing a new feature representation.•A data augmentation strategy to effectively leverage the relationship of the compounds in the training set.•Design a new reward function for the deep reinforcement learning model for feature selection.