李宏毅MLDS课程笔记9:Generative Adversarial Network(GAN)
NIPS 2016 Tutorial: Generative Adversarial Networks [Paper] [Video]
tips for training GAN: https://github.com/soumith/ganhacksBasic Idea of GAN1、最大化似然函数等价于最小化PdataP_{data} 与PGP_G 之间的KL散度。如果PGP_G 取的是GM
原创
2017-04-23 16:01:45 ·
4325 阅读 ·
0 评论