End-to-end approach of multi-grained embedding of categorical features in tabular data

Han Liu, Qianxi Qiu, Qin Zhang

Abstract: Highlights•We propose an approach of multi-grained embedding of features in tabular data.•The proposed approach works in an end-to-end manner with no neural network.•An uncertainty-aware optimization strategy is designed for end-to-end learning.•Our approach achieves learning from categorical data with no need of handcrafting.•The results show the superiority of our approach in comparison with baselines.