Fig. 4From: YuYin: a multi-task learning model of multi-modal e-commerce background music recommendationThe weighted fusion module (WFM) in YuYin, which learns to apply dynamic weights for audio and emotion features and output the music featureBack to article page