Self-supervised GAN: Analysis and Improvement with Multi-class Minimax Game

Authors: Ngoc-Trung Tran, Viet-Hung Tran, Bao-Ngoc Nguyen, Linxiao Yang, Ngai-Man (Man) Cheung

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this work, we perform an in-depth analysis to understand how SS tasks interact with learning of generator. From the analysis, we identify issues of SS tasks which allow a severely mode-collapsed generator to excel the SS tasks. To address the issues, we propose new SS tasks based on a multi-class minimax game. The competition between our proposed SS tasks in the game encourages the generator to learn the data distribution and generate diverse samples. We provide both theoretical and empirical analysis to support that our proposed SS tasks have better convergence property. We conduct experiments to incorporate our proposed SS tasks into two different GAN baseline models. Our approach establishes state-of-the-art FID scores on CIFAR-10, CIFAR-100, STL-10, Celeb A, Imagenet 32 32 and Stacked-MNIST datasets, outperforming existing works by considerable margins in some cases.
Researcher Affiliation Academia Ngoc-Trung Tran, Viet-Hung Tran, Ngoc-Bao Nguyen, Linxiao Yang, Ngai-Man Cheung Singapore University of Technology and Design (SUTD)
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Our code: https://github.com/tntrung/msgan
Open Datasets Yes Our approach establishes state-of-the-art FID scores on CIFAR-10, CIFAR-100, STL-10, Celeb A, Imagenet 32 32 and Stacked-MNIST datasets, outperforming existing works by considerable margins in some cases.
Dataset Splits No The paper refers to using standard datasets but does not explicitly state the specific proportions, methods, or sizes of the training, validation, and test splits needed to reproduce the data partitioning. While it mentions how FID is computed with real and generated samples, it does not detail the dataset split for a validation set.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or cloud instance types) used for running its experiments.
Software Dependencies No The paper mentions software but does not provide specific version numbers for any key software components or libraries.
Experiment Setup Yes We report the best FID attained in 300K iterations as in [45, 25, 42, 47]. We keep all parameters suggested in the original work and focus to understand the contribution of our proposed techniques. For each experiment and for each approach (SS or MS), we obtain the best λg and λd using extensive search (see Appendix B.4 for details), and we use the best λg and λd in the comparison depicted in Fig. 4.