AI helps you reading Science

AI generates interpretation videos

AI extracts and analyses the key points of the paper to generate videos automatically


pub
Go Generating

AI Traceability

AI parses the academic lineage of this thesis


Master Reading Tree
Generate MRT

AI Insight

AI extracts a summary of this paper


Weibo:
We introduce an encoder in a conditional setting within the Generative Adversarial Networks framework, a model which we call Invertible Conditional Generative Adversarial Networks

Invertible Conditional GANs for image editing.

arXiv: Computer Vision and Pattern Recognition, (2016)

Cited: 346|Views180
EI
Full Text
Bibtex
Weibo

Abstract

Generative Adversarial Networks (GANs) have recently demonstrated to successfully approximate complex data distributions. A relevant extension of this model is conditional GANs (cGANs), where the introduction of external information allows to determine specific representations of the generated images. In this work, we evaluate encoders to...More

Code:

Data:

0
Introduction
  • Image editing can be performed at different levels of complexity and abstraction. Common operations consist in applying a filter to an image to, for example, augment the contrast or convert to grayscale.
  • If one would want to modify the attributes of a face, this is a more complex and challenging modification to perform.
  • In this case, in order to obtain realistic results, a skilled human with an image edition software would often be required.
  • A GAN is composed of two neural networks, a generator G and a discriminator D.
Highlights
  • Image editing can be performed at different levels of complexity and abstraction
  • Generative Adversarial Networks (GANs) [3] is one of the state-of-the-art approaches for image generation
  • In order to overcome this limitation, in this paper we introduce Invertible Conditional Generative Adversarial Networks (IcGANs) for complex image editing as the union of an encoder used jointly with a conditional GANs
  • We introduce Invertible Conditional Generative Adversarial Networks (IcGANs), which are composed of a conditional GANs and an encoder
  • We introduce an encoder in a conditional setting within the Generative Adversarial Networks framework, a model which we call Invertible Conditional Generative Adversarial Networks (IcGANs)
  • The results obtained with a complex face dataset, CelebFaces Attributes, are satisfactory and promising
Methods
  • The authors use two image datasets of different complexity and variation, MNIST [5] and CelebFaces Attributes (CelebA) [6].
  • Each sample is a 28 × 28 centered image labeled with the class of the digit (0 to 9).
  • CelebA is a dataset composed of 202,599 face colored images and 40 attribute binary vectors.
  • The authors will evaluate the quality of generated samples of both datasets.
  • A quantitative evaluation will be performed on CelebA only, as it is considerably more complex than MNIST
Conclusion
  • The authors introduce an encoder in a conditional setting within the GAN framework, a model which the authors call Invertible Conditional GANs (IcGANs).
  • It solves the problem of GANs lacking the ability to infer real samples to a latent representation z, while allowing to explicitly control complex attributes of generated samples with conditional information y.
  • The results obtained with a complex face dataset, CelebA, are satisfactory and promising
Tables
  • Table1: Detailed generator and discriminator architecture
  • Table2: Encoder IND architecture. Last two layers have different sizes depending on the encoder (z for Ez or y for Ey). ny represents the size of y
  • Table3: Quantitative cGAN evaluation depending on y inserted position. The first row shows the results obtained with real CelebA images as an indication that Anet predictions are subject to error
Download tables as Excel
Related work
  • There are different approaches for generative models. Among them, there are two promising ones that are recently pushing the state-of-the-art with highly plausible generated images.

    The first one is Variational Autoencoders (VAE) [1, 4, 7, 8], which impose a prior representation space z (e.g. normal distribution) in order to regularize and constrain the model to sample from it. However, VAEs main limitation is the pixel-wise reconstruction error used as a loss function, which causes the output images to look blurry. The second approach is Generative Adversarial Nets (GANs). Originally proposed by Goodfellow et al [3], GANs have been improved with a deeper architecture (DCGAN) by Radford et al [2]. The latest advances introduced several techniques that improve the overall performance for training GANs [9] and an unsupervised approach to disentangle feature representations [10]. Additionally, the most advanced and recent work on cGANs trains a model to generate realistic images from text descriptions and landmarks [11].
Funding
  • This work is funded by the Projects TIN2013-41751-P of the Spanish Ministry of Science and the CHIST ERA project PCIN-2015-226
Study subjects and analysis
image datasets: 2
The combination of an encoder with a cGAN, which we call Invertible cGAN (IcGAN), enables to re-generate real images with deterministic complex modifications. 5.1 Datasets

We use two image datasets of different complexity and variation, MNIST [5] and CelebFaces Attributes (CelebA) [6]
. MNIST is a digit dataset of grayscale images composed of 60,000 training images and 10,000 test images

image datasets: 2
5.1 Datasets. We use two image datasets of different complexity and variation, MNIST [5] and CelebFaces Attributes (CelebA) [6]. MNIST is a digit dataset of grayscale images composed of 60,000 training images and 10,000 test images

Reference
  • K. Gregor, I. Danihelka, A. Graves, D. Jimenez Rezende, and D. Wierstra, “DRAW: A Recurrent Neural Network For Image Generation,” International Conference on Machine Learning (ICML), pp. 1462–1471, 2015.
    Google ScholarLocate open access versionFindings
  • A. Radford, L. Metz, and S. Chintala, “Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks,” International Conference on Learning Representations (ICLR), 2016.
    Google ScholarLocate open access versionFindings
  • I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative adversarial nets,” in Advances in Neural Information Processing Systems 27, Z. Ghahramani, M. Welling, C. Cortes, N. D. Lawrence, and K. Q. Weinberger, Eds. Curran Associates, Inc., 2014, pp. 2672–2680. [Online]. Available: http://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf
    Findings
  • D. P. Kingma and M. Welling, “Auto-encoding variational bayes,” International Conference on Learning Representations (ICLR), 201[Online]. Available: http://arxiv.org/abs/1312.6114
    Findings
  • Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” in Proceedings of the IEEE, 1998, pp. 2278–2324.
    Google ScholarLocate open access versionFindings
  • Z. Liu, P. Luo, X. Wang, and X. Tang, “Deep learning face attributes in the wild,” in Proceedings of International Conference on Computer Vision (ICCV), December 2015.
    Google ScholarLocate open access versionFindings
  • D. J. Rezende, S. Mohamed, and D. Wierstra, “Stochastic backpropagation and approximate inference in deep generative models,” Proceedings of the 31st International Conference on Machine Learning (ICML), vol. 32, pp. 1278–1286, 2014.
    Google ScholarLocate open access versionFindings
  • D. P. Kingma, D. J. Rezende, S. Mohamed, and M. Welling, “Semi-supervised learning with deep generative models,” Proceedings of Neural Information Processing Systems (NIPS), 2014. [Online]. Available: http://arxiv.org/abs/1406.5298
    Findings
  • T. Salimans, I. J. Goodfellow, W. Zaremba, V. Cheung, A. Radford, and X. Chen, “Improved techniques for training GANs,” Neural Information Processing Systems (NIPS), 2016. [Online]. Available: http://arxiv.org/abs/1606.03498
    Findings
  • X. Chen, Y. Duan, R. Houthooft, J. Schulman, I. Sutskever, and P. Abbeel, “InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets,” Neural Information Processing Systems (NIPS), 2016. [Online]. Available: http://arxiv.org/abs/1606.03657
    Findings
  • S. Reed, Z. Akata, S. Mohan, S. Tenka, B. Schiele, and H. Lee, “Learning what and where to draw,” in Advances in Neural Information Processing Systems (NIPS), 2016.
    Google ScholarLocate open access versionFindings
  • V. Dumoulin, I. Belghazi, B. Poole, A. Lamb, M. Arjovsky, O. Mastropietro, and A. Courville, “Adversarially Learned Inference,” Arxiv, 2016. [Online]. Available: http://arxiv.org/abs/1606.00704
    Findings
  • J. Donahue, P. Krähenbühl, and T. Darrell, “Adversarial feature learning,” CoRR, vol. abs/1605.09782, 2016. [Online]. Available: http://arxiv.org/abs/1605.09782
    Findings
  • A. Makhzani, J. Shlens, N. Jaitly, and I. Goodfellow, “Adversarial autoencoders,” in International Conference on Learning Representations (ICLR), 2016. [Online]. Available: http://arxiv.org/abs/1511.05644
    Findings
  • A. B. L. Larsen, S. K. Sønderby, and O. Winther, “Autoencoding beyond pixels using a learned similarity metric,” Proceedings of the 33rd International Conference on Machine Learning (ICML), pp. 1558–1566, 20[Online]. Available: http://arxiv.org/abs/1512.09300
    Findings
  • S. E. Reed, Z. Akata, X. Yan, L. Logeswaran, B. Schiele, and H. Lee, “Generative adversarial text to image synthesis,” International Conference on Machine Learning (ICML), 20[Online]. Available: http://arxiv.org/abs/1605.05396
    Findings
  • M. Mirza and S. Osindero, “Conditional Generative Adversarial Nets,” CoRR, vol. abs/1411.1784, 2014. [Online]. Available: http://arxiv.org/abs/1411.1784
    Findings
  • J. Gauthier, “Conditional generative adversarial nets for convolutional face generation,” Class project for Stanford CS231N: Convolutional Neural Networks for Visual Recognition, Winter semester 2014, 2014. [Online]. Available: http://cs231n.stanford.edu/reports/jgauthie_final_report.pdf
    Findings
  • D. Kingma and J. Ba, “Adam: A Method for Stochastic Optimization,” International Conference for Learning Representations (ICLR), 2014. [Online]. Available: http://www.arxiv.org/pdf/1412.6980.pdf
    Findings
0
Your rating :

No Ratings

Tags
Comments
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn