Skip to main content
SpringerLink
Log in

DEGAS: differentiable efficient generator search

  • Original Article
  • Published:
Neural Computing and Applications Aims and scope Submit manuscript

Abstract

Network architecture search achieves state-of-the-art results in various tasks such as classification and semantic segmentation. Recently, a reinforcement learning-based approach has been proposed for generative adversarial networks (GANs) search. In this work, we propose an alternative strategy for GAN search by using a proxy task instead of common GAN training. Our method is called differentiable efficient generator search, which focuses on efficiently finding the generator in the GAN. Our search algorithm is inspired by the differential architecture search strategy and the global latent optimization procedure. This leads to both an efficient and stable GAN search. After the generator architecture is found, it can be plugged into any existing framework for GAN training. For consistency-term GAN, which we use in this work, the new model outperforms the original inception score results by 0.25 for CIFAR-10.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7

Similar content being viewed by others

Availability of data and material

Open-source data were used in this paper

References

  1. Abu Hussein S, Tirer T, Giryes R (2020) Image-adaptive GAN based reconstruction. In: AAAI conference on artificial intelligence

  2. Andrychowicz M, Denil M, Gomez Colmenarejo S, Hoffman MW, Pfau D, Schaul T, Shillingford B, de Freitas N (2016) Learning to learn by gradient descent by gradient descent. In: Advances in neural information processing systems (NIPS)

  3. Arjovsky M, Chintala S, Bottou L (2017) Wasserstein gan. arXiv:1701.07875

  4. Bojanowski P, Joulin A, Lopez Paz D, Szlam A (2018) Optimizing the latent space of generative networks. In: International conference on machine learning (ICML)

  5. Bora A, Jalal A, Price E, Dimakis AG (2017) Compressed sensing using generative models. PMLR, International Convention Centre, Sydney, Australia, pp 537–546

  6. Brock A, Donahue J, Simonyan K (2018) Large scale gan training for high fidelity natural image synthesis. arXiv:1809.11096v2

  7. Cai H, Zhu L, Han S (2019) Proxylessnas: direct neural architecture search on target task and hardware. In: ICLR

  8. Chen X, Xie L, Wu J, Tian Q (2019) Progressive differentiable architecture search: Bridging the depth gap between search and evaluation. arXiv:1904.12760

  9. Chen Y, Hoffman MW, Colmenarejo SG, Denil M, Lillicrap TP, Botvinick M, de Freitas N (2017) Learning to learn without gradient descent by gradient descent. In: International conference on machine learning (ICML)

  10. Clark A, Donahue J, Simonyan K (2019) Adversarial video generation on complex datasets. arXiv:1907.06571

  11. Coates A, Ng A, Lee H (2011) An analysis of single-layer networks in unsupervised feature learning. In: Proceedings of the fourteenth international conference on artificial intelligence and statistics. JMLR workshop and conference proceedings, pp 215–223

  12. Cubuk ED, Zoph B, Mane D, Vasudevan V, Viet Le Q (2018) Autoaugment: Learning augmentation policies from data. arXiv:1805.09501v2

  13. Feigin Y, Spitzer H, Giryes R (2020) Gmm-based generative adversarial encoder learning

  14. Geifman Y, El-Yaniv R (2019) Deep active learning with a neural architecture search. In: NeurIPS

  15. Gong X, Chang S, Jiang Y, Wang Z (2019) Autogan: Neural architecture search for generative adversarial networks. In: ICCV

  16. Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y (2014) Generative adversarial nets. In: Advances in neural information processing systems, pp 2672–2680

  17. Gulrajani I, Ahmed F, Arjovsky M, Dumoulin V, Courville A (2017) Improved training of wasserstein gans. arXiv:1704.00028

  18. He K, Zhang X, Ren S, Sun J (2015) Deep residual learning for image recognition. arXiv:1512.03385

  19. Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S (2017) Gans trained by a two time-scale update rule converge to a local nash equilibrium. In: Advances in neural information

  20. Hoshen Y, Wolf L (2018) Nam: Non-adversarial unsupervised domain mapping. In: ECCV, pp 455–470

  21. Huang Y, Cheng Y, Chen D, Lee H, Ngiam J, Viet Le Q, Chen, Z (2018) Gpipe: Efficient training of giant neural networks using pipeline parallelism. arXiv:1811.06965

  22. Isola P, Zhu JY, Zhou T, Efros AA (2017) Image-to-image translation with conditional adversarial networks. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 1125–1134

  23. Kaplan S, Giryes R (2020) Self-supervised neural architecture search

  24. Karras T, Aila T, Laine S, Lehtinen J (2017) Progressive growing of gans for improved quality, stability, and variation. In: ICLR

  25. Karras T, Laine S, Aila T (2019) A style-based generator architecture for generative adversarial networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 4401–4410

  26. Krizhevsky A, Hinton G (2009) Learning multiple layers of features from tiny images. University of Toronto, Toronto

    Google Scholar 

  27. Kurach K, Lucic M, Zhai X, Michalski M, Gelly S (2019) A large-scale study on regularization and normalization in gans. Int Conf Mach Learn 97:3581–3590

    Google Scholar 

  28. Liu C, Chen LC, Schroff F, Adam H, Hua W, Yuille A, Li FF (2019) Auto-deeplab: hierarchical neural architecture search for semantic image segmentation. In: CVPR

  29. Liu C, Dollár P, He K, Girshick R, Yuille A, Xie S (2020) Are labels necessary for neural architecture search? In: ECCV

  30. Liu H, Simonyan K, Yang Y (2019) Darts: differentiable architecture search. In: International conference on learning representations (ICLR)

  31. Liu Z, Luo P, Wang X, Tang X (2018) Large-scale celebfaces attributes (celeba) dataset. Retrieved August 15(2018):11

  32. Miyato T, Kataoka T, Koyama M, Yoshida Y (2018) Spectral normalization for generative adversarial networks. In: ICLR

  33. Ni Y, Song D, Zhang X, Wu H, Liao L (2018) Cagan: Consistent adversarial training enhanced gans. In: International joint conference on artificial intelligence (IJCAI)

  34. Ning W, Yang G, Hao C, Peng W, Zhi T, Chunhua S (2019) Nas-fcos: Fast neural architecture search for object detection. arXiv:1906.04423

  35. Noy A, Nayman N, Ridnik T, Zamir N, Doveh S, Friedman I, Giryes R, Zelnik-Manor L (2019) Asap: architecture search, anneal and prune. arXiv:1904.04123

  36. Odena A, Olah C, Shlens J (2016) Conditional image synthesis with auxiliary classifier gans. In: ICML

  37. Pham H, Guan MY, Zoph B, Viet Le Q, Dean J (2018) Efficient neural architecture search via parameter sharing. In: International conference on machine learning (ICML)

  38. Radford A, Metz L, Chintala S (2016) Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv:1511.06434

  39. Raj A, Li Y, Bresler Y (2019) Gan-based projector for faster recovery with convergence guarantees in linear inverse problems. In: 2019 IEEE/CVF international conference on computer vision (ICCV), pp 5601–5610. https://doi.org/10.1109/ICCV.2019.00570

  40. Real E, Aggarwal A, Huang Y, Viet Le Q (2018) Regularized evolution for image classifier architecture search. In: International conference on machine learning-ICML AutoML workshop

  41. Real E, Moore S, Selle A, Saxena S, Suematsu YL, Tan J, Viet Le Q, Kurakin A (2017) Large-scale evolution of image classifiers. In: International conference on machine learning (ICML)

  42. Rubin M, Stein O, Turko NA, Nygate Y, Roitshtain D, Karako L, Barnea I, Giryes R, Shaked NT (2019) Top-gan: stain-free cancer cell classification using deep learning with a small training set. Med Image Anal 57:176–185

    Article  Google Scholar 

  43. Salimans T, Goodfellow I, Zaremba W, Cheung V, Radford A, Chen X (2016) Improved techniques for training gans. In: Advances in neural information processing systems, pp 2234–2242

  44. Sun G, Ding S, Sun T, Zhang C (2021) SA-CapsGAN: using capsule networks with embedded self-attention for generative adversarial network. Neurocomputing 423:399–406. https://doi.org/10.1016/j.neucom.2020.10.092

    Article  Google Scholar 

  45. Tulyakov S, Liu MY, Yang X, Kautz J (2018) Mocogan: decomposing motion and content for video generation. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 1526–1535

  46. Wang H, Huan J (2019) Agan: towards automated design of generative adversarial networks. arXiv:1906.11080

  47. Wei X, Gong B, Liu Z, Lu W, Wang L (2018) Improving the improved training of wasserstein gans: a consistency term and its dual effect. In: International conference on learning representation (ICLR)

  48. Weng Y, Zhou T, Li Y, Qiu X (2019) Nas-unet: neural architecture search for medical image segmentation. In: IEEE Access, vol 7

  49. Yan S, Zheng Y, Ao W, Zeng X, Zhang M (2020) Does unsupervised architecture representation learning help neural architecture search? NeurIPS 33

  50. Zhang J, Ding S, Zhang N, Jia W (2020) Adversarial training methods for Boltzmann machines. IEEE Access 8:4594–4604. https://doi.org/10.1109/ACCESS.2019.2962758

    Article  Google Scholar 

  51. Zhu JY, Park T, Isola P, Efros AA (2017) Unpaired image-to-image translation using cycle-consistent adversarial networks. In: Proceedings of the IEEE international conference on computer vision, pp 2223–2232

  52. Zhu Z, Liu C, Yang D, Yuille A, Xu D (2019) V-nas: neural architecture search for volumetric medical image segmentation. arXiv:1906.0281

  53. Zoph B, Viet Le Q (2017) Neural architecture search with reinforcement learning. In: International conference on learning representations (ICLR)

Download references

Acknowledgements

This work was supported by Alibaba and the NSF-BSF grant.

Author information

Authors and Affiliations

  1. Department of Electrical Engineering, Tel Aviv University, 69978, Tel Aviv, Israel

    Sivan Doveh & Raja Giryes

Authors
  1. Sivan Doveh
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Raja Giryes
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Sivan Doveh.

Ethics declarations

Conflict of interest

IBM-Research intern.

Code availability

Code will be released upon acceptance.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Doveh, S., Giryes, R. DEGAS: differentiable efficient generator search. Neural Comput & Applic 33, 17173–17184 (2021). https://doi.org/10.1007/s00521-021-06309-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00521-021-06309-8

Keywords

Search

Navigation