TY - GEN
T1 - ON the tradeoff between mode collapse and sample quality in generative adversarial networks
AU - Adiga, Sudarshan
AU - Attia, Mohamed Adel
AU - Chang, Wei Ting
AU - Tandon, Ravi
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/7/2
Y1 - 2018/7/2
N2 - Generative Adversarial Networks (GAN) are used to generate synthetic samples while closely following the underlying distribution of a real data set. While GANs have recently gained significant popularity, they often suffer from the mode collapse problem, where the generated samples lack diversity. Moreover, some approaches that attempt to resolve the model collapse problem do not necessarily yield high quality synthetic samples. In this paper, we propose two novel performance metrics, namely mode-collapse divergence (MCD) which quantifies the extent of mode collapse for a GAN architecture. Second, we propose the metric Generative Quality Score (GQS), which measures the quality of generated samples. We present a comprehensive study of the performance of various GAN architectures proposed in the literature through the lens of MCD and GQS, for three different data sets, namely MNIST, Fashion MNIST and CIFAR-10.
AB - Generative Adversarial Networks (GAN) are used to generate synthetic samples while closely following the underlying distribution of a real data set. While GANs have recently gained significant popularity, they often suffer from the mode collapse problem, where the generated samples lack diversity. Moreover, some approaches that attempt to resolve the model collapse problem do not necessarily yield high quality synthetic samples. In this paper, we propose two novel performance metrics, namely mode-collapse divergence (MCD) which quantifies the extent of mode collapse for a GAN architecture. Second, we propose the metric Generative Quality Score (GQS), which measures the quality of generated samples. We present a comprehensive study of the performance of various GAN architectures proposed in the literature through the lens of MCD and GQS, for three different data sets, namely MNIST, Fashion MNIST and CIFAR-10.
KW - Generative Adversarial Networks
KW - Mode Collapse
KW - Performance Metrics
UR - http://www.scopus.com/inward/record.url?scp=85063093478&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85063093478&partnerID=8YFLogxK
U2 - 10.1109/GlobalSIP.2018.8646478
DO - 10.1109/GlobalSIP.2018.8646478
M3 - Conference contribution
AN - SCOPUS:85063093478
T3 - 2018 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2018 - Proceedings
SP - 1184
EP - 1188
BT - 2018 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2018 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2018 IEEE Global Conference on Signal and Information Processing, GlobalSIP 2018
Y2 - 26 November 2018 through 29 November 2018
ER -