Multi-Time Scale Aware Host Task Preferred Learning for WEEE return prediction

Yujiang Wu, Min Gao, Ruiqi Liu, Jie Zeng, Quanwu Zhao, Jinyong Gao, Jia Zhang

Abstract: Highlights•The relationships among various types of WEEE are used for RL return prediction.•Multi-time scale features and multi-task common features are extracted collaboratively.•Polling host-task learning strategy is proposed to achieve fair prediction.•Host-preferred loss is designed to ensure the highest learning speed of host task.•Sharing tower is added to avoid overly depending on some time series features.