Diffusion Autoencoders: Toward a Meaningful and Decodable Representation
DiffusionAutoencoders:TowardaMeaningfulandDecodableRepresentation(Paperreading)KonpatPreechakul,VISTEC,Thailand,CVPR22Oral,Cited:117,Code,Paper1.前言扩散概率模型(DPM)在图像生成方面取得了显着的质量,可与GAN相媲美。但是与GAN不同,DPM使用一组潜