I. Introduction
Deep learning penetrates the metaverse industry with vigorous modeling capability, such as creating virtual avatars. However, to improve the vividness of virtual avatars, deep learning models’ training requires mountainous real-world human behavioral data, such as appearance, movements, expressions, and speech, which are highly sensitive for their creators. With a series of regulations, such as the General Data Protection Regulation (GDPR), being proposed to protect personal training data, the metaverse industry should pay particular attention to training data privacy.