Evolutionary Architecture Search for Generative Adversarial Networks Based on Weight Sharing

Xue, Yu; Tong, Weinan; Neri, Ferrante; Chen, Peng; Luo, Tao; Zhen, Liangli; Wang, Xiao

doi:10.1109/tevc.2023.3338371

Evolutionary Architecture Search for Generative Adversarial Networks Based on Weight Sharing

Journal Article · Fri Dec 01 00:00:00 EST 2023 · IEEE Transactions on Evolutionary Computation

DOI:https://doi.org/10.1109/tevc.2023.3338371· OSTI ID:2429864

^[1]; Tong, Weinan ^[1]; ^[2]; ^[3]; ^[4]; ^[4]; ^[5]

Nanjing Univ. of Information Science and Technology (China)
Nanjing Univ. of Information Science and Technology (China); Univ. of Surrey, Guildford (United Kingdom)
National Institute of Advanced Industrial Science and Technology (AIST), Tokyo (Japan); RIKEN Center for Computational Science, Kobe (Japan)
Agency for Science, Technology and Research (A*STAR), Fusionopolis (Singapore)
Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)

Generative adversarial networks (GANs) are a powerful generative technique but frequently face challenges with training stability. Network architecture plays a significant role in determining the final output of GANs, but designing a fine architecture demands extensive domain expertise. This article aims to address this issue by searching for high-performance generator’s architectures through neural architecture search (NAS). The proposed approach, called evolutionary weight sharing GANs (EWSGAN), is based on weight sharing and comprises two steps. First, a supernet of the generator is trained using weight sharing. Second, a multiobjective evolutionary algorithm (MOEA) is employed to identify optimal subnets from the supernet. These subnets inherit weights directly from the supernet for fitness assessment. Two strategies are used to stabilize the training of the generator supernet: 1) a fair single-path sampling strategy and 2) a discarding strategy. Experimental results indicate that the architecture searched by our method achieved a new state-of-the-art among NAS–GAN methods with a Fréchet inception distance (FID) of 9.09 and an inception score (IS) of 8.99 on the CIFAR-10 dataset. Finally, it also demonstrates competitive performance on the STL-10 dataset, achieving FID of 21.89 and IS of 10.51.

View Accepted Manuscript (DOE)

Research Organization:: Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)

Sponsoring Organization:: National Natural Science Foundation of China (NSFC); Natural Science Foundation of Jiangsu Province; Natural Science Foundation of the Jiangsu Higher Education Institutions of China; USDOE

Grant/Contract Number:: AC05-00OR22725

OSTI ID:: 2429864

Journal Information:: IEEE Transactions on Evolutionary Computation, Journal Name: IEEE Transactions on Evolutionary Computation Journal Issue: 3 Vol. 28; ISSN 1089-778X

Publisher:: IEEECopyright Statement

Country of Publication:: United States

Language:: English

References (32)

DEGAS: differentiable efficient generator search Doveh, Sivan; Giryes, Raja Neural Computing and Applications, Vol. 33, Issue 24 https://doi.org/10.1007/s00521-021-06309-8	journal	July 2021
Analyze COVID-19 CT images based on evolutionary algorithm with dynamic searching space Gong, Yunhong; Sun, Yanan; Peng, Dezhong Complex & Intelligent Systems, Vol. 7, Issue 6 https://doi.org/10.1007/s40747-021-00513-8	journal	September 2021
Pros and cons of GAN evaluation measures: New developments Borji, Ali Computer Vision and Image Understanding, Vol. 215 https://doi.org/10.1016/j.cviu.2021.103329	journal	January 2022
A fast and elitist multiobjective genetic algorithm: NSGA-II Deb, K.; Pratap, A.; Agarwal, S. IEEE Transactions on Evolutionary Computation, Vol. 6, Issue 2 https://doi.org/10.1109/4235.996017	journal	April 2002
Deep Residual Learning for Image Recognition He, Kaiming; Zhang, Xiangyu; Ren, Shaoqing 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2016.90	conference	June 2016
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network Ledig, Christian; Theis, Lucas; Huszar, Ferenc 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2017.19	conference	July 2017
Perceptual Generative Adversarial Networks for Small Object Detection Li, Jianan; Liang, Xiaodan; Wei, Yunchao 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2017.211	conference	July 2017
Learning Transferable Architectures for Scalable Image Recognition Zoph, Barret; Vasudevan, Vijay; Shlens, Jonathon 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2018.00907	conference	June 2018
A Style-Based Generator Architecture for Generative Adversarial Networks Karras, Tero; Laine, Samuli; Aila, Timo 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR.2019.00453	conference	June 2019
GreedyNAS: Towards Fast One-Shot NAS With Greedy Supernet You, Shan; Huang, Tao; Yang, Mingmin 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00207	conference	June 2020
GAN Compression: Efficient Architectures for Interactive Conditional GANs Li, Muyang; Lin, Ji; Ding, Yaoyao 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00533	conference	June 2020
AdversarialNAS: Adversarial Neural Architecture Search for GANs Gao, Chen; Chen, Yunpeng; Liu, Si 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00572	conference	June 2020
Analyzing and Improving the Image Quality of StyleGAN Karras, Tero; Laine, Samuli; Aittala, Miika 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.00813	conference	June 2020
EcoNAS: Finding Proxies for Economical Neural Architecture Search Zhou, Dongzhan; Zhou, Xinchi; Zhang, Wenwei 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) https://doi.org/10.1109/CVPR42600.2020.01141	conference	June 2020
Deep Learning Face Attributes in the Wild Liu, Ziwei; Luo, Ping; Wang, Xiaogang 2015 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2015.425	conference	December 2015
Least Squares Generative Adversarial Networks Mao, Xudong; Li, Qing; Xie, Haoran 2017 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2017.304	conference	October 2017
StackGAN: Text to Photo-Realistic Image Synthesis with Stacked Generative Adversarial Networks Zhang, Han; Xu, Tao; Li, Hongsheng 2017 IEEE International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2017.629	conference	October 2017
AutoGAN: Neural Architecture Search for Generative Adversarial Networks Gong, Xinyu; Chang, Shiyu; Jiang, Yifan 2019 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV.2019.00332	conference	October 2019
FairNAS: Rethinking Evaluation Fairness of Weight Sharing Neural Architecture Search Chu, Xiangxiang; Zhang, Bo; Xu, Ruijun 2021 IEEE/CVF International Conference on Computer Vision (ICCV) https://doi.org/10.1109/ICCV48922.2021.01202	conference	October 2021
Automatically Designing CNN Architectures Using the Genetic Algorithm for Image Classification Sun, Yanan; Xue, Bing; Zhang, Mengjie IEEE Transactions on Cybernetics, Vol. 50, Issue 9 https://doi.org/10.1109/TCYB.2020.2983860	journal	September 2020
Evolving Deep Convolutional Neural Networks for Image Classification Sun, Yanan; Xue, Bing; Zhang, Mengjie IEEE Transactions on Evolutionary Computation, Vol. 24, Issue 2 https://doi.org/10.1109/TEVC.2019.2916183	journal	April 2020
A Survey on Evolutionary Neural Architecture Search Liu, Yuqiao; Sun, Yanan; Xue, Bing IEEE Transactions on Neural Networks and Learning Systems, Vol. 34, Issue 2 https://doi.org/10.1109/TNNLS.2021.3100554	journal	February 2023
AlphaGAN: Fully Differentiable Architecture Search for Generative Adversarial Networks Tian, Yuesong; Shen, Li; Shen, Li IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 44, Issue 10 https://doi.org/10.1109/TPAMI.2021.3099829	journal	October 2022
NSGA-Net Lu, Zhichao; Whalen, Ian; Boddeti, Vishnu Proceedings of the Genetic and Evolutionary Computation Conference https://doi.org/10.1145/3321707.3321729	conference	July 2019
Coegan Costa, Victor; Lourenço, Nuno; Correia, João Proceedings of the Genetic and Evolutionary Computation Conference https://doi.org/10.1145/3321707.3321746	conference	July 2019
Dynamically Grown Generative Adversarial Networks Liu, Lanlan; Zhang, Yuting; Deng, Jia Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35, Issue 10 https://doi.org/10.1609/aaai.v35i10.17052	journal	May 2021
Estimates of the Regression Coefficient Based on Kendall's Tau Sen, Pranab Kumar Journal of the American Statistical Association, Vol. 63, Issue 324 https://doi.org/10.2307/2285891	journal	December 1968
Individual Comparisons by Ranking Methods Wilcoxon, Frank Biometrics Bulletin, Vol. 1, Issue 6 https://doi.org/10.2307/3001968	journal	December 1945
Neural Architecture Search with Reinforcement Learning Zoph, Barret; Le, Quoc V. arXiv https://doi.org/10.48550/ARXIV.1611.01578	preprint	January 2016
Attention Is All You Need Vaswani, Ashish; Shazeer, Noam; Parmar, Niki arXiv https://doi.org/10.48550/ARXIV.1706.03762	preprint	January 2017
Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks Radford, Alec; Metz, Luke; Chintala, Soumith arXiv https://doi.org/10.48550/arXiv.1511.06434	preprint	January 2015
Spectral Normalization for Generative Adversarial Networks Miyato, Takeru; Kataoka, Toshiki; Koyama, Masanori arXiv https://doi.org/10.48550/arXiv.1802.05957	preprint	January 2018

Similar Records

Stable parallel training of Wasserstein conditional generative adversarial neural networks

Journal Article · Tue Aug 02 20:00:00 EDT 2022 · Journal of Supercomputing · OSTI ID:1908079

Stable Parallel Training of Wasserstein Conditional Generative Adversarial Neural Networks : *Full/Regular Research Paper submission for the symposium CSCI-ISAI: Artificial Intelligence

Conference · Tue Nov 30 23:00:00 EST 2021 · OSTI ID:1877492

Scalable balanced training of conditional generative adversarial neural networks on image data

Journal Article · Sun Apr 25 20:00:00 EDT 2021 · Journal of Supercomputing · OSTI ID:1783019

Related Subjects

97 MATHEMATICS AND COMPUTING
evolutionary computation
generative adversarial networks (GANs)
generative model
neural architecture search (NAS)

Evolutionary Architecture Search for Generative Adversarial Networks Based on Weight Sharing

Citation Formats

References (32)

Similar Records

Related Subjects