Failure Modes in Unsupervised Image-to-Image Translation
Main references:
- “Generalization and Equilibrium in Generative Adversarial Nets” by Arora et al., PMLR 2017.
- “Training Generative Adversarial Networks with Limited Data” by Karras et al., NeurIPS 2020.
- “Risk Bounds for Unsupervised Cross-Domain Mapping with IPMs” by Galanti et al., JMLR 2021. // 2017-2021
Other references:
“Simple Strategies for Large Zero-Sum Games with Applications to Complexity Theory” by Lipton & Young, STOC’94
“Foundations of Machine Learning” Mohri, Rostamizadeh, Talwalkar, 2nd Edition, 2018
“Towards Principled Methods for Training Generative Adversarial Networks”, Arjovsky & Bottou, ICLR’17
“Stabilizing Training of Generative Adversarial Networks through Regularization”, Roth et al, NeurIPS’17
“Which Training Methods for GANs do actually Converge?”, Mescheder et al., ICML’18
“Kernel of CycleGAN as a Principle homogeneous space”, Moriakov et al., ICLR’20
“Guiding the One-to-One Mapping in CycleGAN via Optimal Transport’, Lu et al., AAAI’19