Tackling mode collapse in multi-generator GANs with orthogonal vectors
作者：
Highlights：
• This paper proposes a novel MGO-GAN model which learns a mapping function parameterized by multiple generators from the randomized space to the original data space, overcoming the problem of mode collapse.
• This paper utilizes the back propagation to minimize the orthogonal value in GAN and combine the orthogonal value with the generator loss to jointly update the parameters of generator from both theoretical and empirical perspectives, offering new insights into the success of MGO-GAN.
• Through comprehensive experiments on three datasets with different resolutions, we demonstrate the effectiveness of the proposed approach.
摘要
•This paper proposes a novel MGO-GAN model which learns a mapping function parameterized by multiple generators from the randomized space to the original data space, overcoming the problem of mode collapse.•This paper utilizes the back propagation to minimize the orthogonal value in GAN and combine the orthogonal value with the generator loss to jointly update the parameters of generator from both theoretical and empirical perspectives, offering new insights into the success of MGO-GAN.•Through comprehensive experiments on three datasets with different resolutions, we demonstrate the effectiveness of the proposed approach.
论文关键词：GANs,Mode collapse,Multiple generators,Orthogonal vectors,Minimax formula
论文评审过程：Received 28 February 2020, Revised 8 July 2020, Accepted 6 September 2020, Available online 14 September 2020, Version of Record 18 September 2020.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107646
