Abstract
Domain adaption is definitely critical for success in bridging source and target domains that data distribution shifts exist in domain or task. The state-of-the-art of the adversarial feature learning model named Bidirectional Generative Adversarial Networks (BiGAN), forces generative models to align with an arbitrarily complex distribution in a latent space. However, BiGAN only matches single data distribution without exploiting multi-domain structure, which means the learned latent representation could not transfer to related target domains. Recent research has proved that GANs combined with Cycle Consistent Constraints are effective at image translation. Therefore, we propose a novel framework named Transferable Bidirectional Generative Adversarial Networks combining with Cycle-Consistent Constraints (Cycle-TBiGAN) be applied in cross-domain translation, which aims at learning an alignment latent feature representation and achieving a mapping function between domains. Our framework is suitable for a wide variety of domain adaption scenarios. We show the surprising results in the task of image translation without prior ground-truth knowledge. Extensive experiments are presented on several public datasets. Quantitative comparisons demonstrate the superiority of our approach against previous methods.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsReferences
Bengio, Y., Courville, A.C., Vincent, P.: Representation learning: a review and new perspectives. IEEE Trans. Pattern Anal. Mach. Intell. 35, 1798–1828 (2013). https://doi.org/10.1109/TPAMI.2013.50
Beutel, A., Chen, J., Zhao, Z., Hsin Chi, E.H.: Data decisions and theoretical implications when adversarially learning fair representations. CoRR abs/1707.00075 (2017). https://arxiv.org/abs/1707.00075
Cao, B., Pan, S.J., Zhang, Y., Yeung, D.Y., Yang, Q.: Adaptive transfer learning. In: AAAI (2010). https://www.ntu.edu.sg/home/sinnopan/publications/[AAAI10]Adaptive%20Transfer%20Learning.pdf
Cao, Y., Long, M., Wang, J.: Unsupervised domain adaptation with distribution matching machines (2018). https://aaai.org/ocs/index.php/AAAI/AAAI18/paper/view/17187
Cao, Z., Long, M., Huang, C., Wang, J.: Transfer adversarial hashing for hamming space retrieval. In: AAAI (2018). https://www.aaai.org/ocs/index.php/AAAI/AAAI18/paper/viewPaper/17256
Che, T., Li, Y., Jacob, A.P., Bengio, Y., Li, W.: Mode regularized generative adversarial networks. CoRR abs/1612.02136 (2016). https://arxiv.org/abs/1612.02136
Dai, W., Yang, Q., Xue, G., Yu, Y.: Boosting for transfer learning. In: AAAI 2010 Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence, vol. 8, pp. 193–200 (2007). https://doi.org/10.1145/1273496.1273521
Donahue, J., Krähenbühl, P., Darrell, T.: Adversarial feature learning. ArXiv abs/1605.09782 (2016). https://arxiv.org/abs/1605.09782
Eric Eaton, M.d.: Selective transfer between learning tasks using task-based boosting. In: Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence, pp. 337–342 (2011). https://dl.acm.org/citation.cfm?id=2900476
Gao, J., Fan, W., Jiang, J., Han, J.: Knowledge transfer via multiple model local structure mapping. In: ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 283–291 (2008). https://doi.org/10.1145/1401890.1401928
Goodfellow, I., et al.: Generative adversarial nets. In: Ghahramani, Z., Welling, M., Cortes, C., Lawrence, N.D., Weinberger, K.Q. (eds.) Advances in Neural Information Processing Systems, vol. 27, pp. 2672–2680. Curran Associates, Inc. (2014). http://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf
Hoshen, Y., Wolf, L.: Nam: Non-adversarial unsupervised domain mapping (2018). https://doi.org/10.1007/978-3-030-01264-9_27
Huayan Wang, Q.Y.: Transfer learning by structural analogy. In: AAAI 2011 Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence, vol. 6, 513–518 (2011). https://dl.acm.org/citation.cfm?id=2900505
Long, M., Wang, J., Ding, G., Shen, D., Yang, Q.: Transfer learning with graph co-regularization. IEEE Trans. Knowl. Data Eng. 26, 1805–1818 (2012). https://doi.org/10.1109/TKDE.2013.97
Long, M., Wang, J., Sun, J.G., Yu, P.S.: Domain invariant transfer kernel learning. IEEE Trans. Knowl. Data Eng. 27, 1519–1532 (2015). https://doi.org/10.1109/TKDE.2014.2373376
Raina, R., Battle, A., Lee, H., Packer,B., Ng, A.: Self-taught learning: transfer learning from unlabeled data. In: Proceedings of the 24th International Conference on Machine Learning, vol. 8, no. 227, pp. 759–766 (2007). https://doi.org/10.1145/1273496.1273592
Tran, N.T., Bui, T.A., Cheung, N.M.: Dist-gan: An improved gan using distance constraints. In: ECCV (2018). https://doi.org/10.1007/978-3-030-01264-9_23
Vapnik, V.: An overview of statistical learning theory. IEEE Trans. Neural Networks 10(5), 988–999 (1999). https://doi.org/10.1109/72.788640
Volpi, R., Morerio, P., Savarese, S., Murino, V.: Adversarial feature augmentation for unsupervised domain adaptation, pp. 5495–5504 (2017). https://doi.org/10.1109/cvpr.2018.00576
Pan, W., Xiang, E.W., Liu, N.N., Yang, Q.: Transfer learning in collaborative filtering for sparsity reduction. In: Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence, pp. 230–235 (2010). https://dl.acm.org/citation.cfm?id=2898644
Pan, W., Xiang, E.W., Yang, Q.: Transfer learning in collaborative filtering with uncertain ratings. In: AAAI 2012 Proceedings of the Twenty-Sixth AAAI Conference on Artificial Intelligence, pp. 662–668 (2012). https://dl.acm.org/citation.cfm?id=2900823
Xu, Y., et al.: A unified framework for metric transfer learning. IEEE Trans. Knowl. Data Eng. 29, 1158–1171 (2017). https://doi.org/10.1109/TKDE.2017.2669193
Zhou, J.T., Pan, S.J., Tsang, I.W., Ho, S.S.: Transfer learning for cross-language text categorization through active correspondences construction. In: AAAI, pp. 2400–2406 (2016). https://dl.acm.org/citation.cfm?id=3016234
Zhu, J.Y., Park, T., Isola, P., Efros, A.A.: Unpaired image-to-image translation using cycle-consistent adversarial networks. In: 2017 IEEE International Conference on Computer Vision (ICCV), pp. 2242–2251 (2017). https://doi.org/10.1109/ICCV.2017.244
Zhu, Y., et al.: Heterogeneous transfer learning for image classification. In: AAAI 2011 Proceedings of the Twenty-Fifth AAAI Conference on Artificial Intelligence, pp. 1304–1309 (2011). https://dl.acm.org/citation.cfm?id=2900630
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
A Supplement for Experiment Implementation
A Supplement for Experiment Implementation
As we have described in Sects. 4.1 and 4.2, all the network architectures are repetition by two blocks, full connected block and convolutional or deconvolutional block, each defined by a fully connected layer at the last top layer, a Batch Normalization layer (BN), and a Dropout layer(P), followed by a fully connected (FC) layer or Convolutional layer (CON) with RELU or Leak_RELU activation functions. The Generator consists of Deconvolution layers (DCON) and the full connected output layers with sigmoid hidden units. Image preprocessing includes linear scaling all image sizes to 28 \(\times \) 28, each image is represented by a 256-dimensional feature vector in feature representation space, which encodes the pixel information of the image.
In this section, we will give a detailed introduction about the specific design used to generate the result presented for Transferable Bidirectional Generative Adversarial Networks (TBiGAN) and Cycle-Consistent TBiGAN. A detailed description of architectures and hyperparameters (learning rate, batch sizes, etc.) is displayed in the following sections. We provide a basic necessary understanding of our experiments.
1.1 A.1 Transferable Bidirectional Generative Adversarial Networks (TBiGAN)
We apply TBiGAN to a task that aims at learning an invariant feature representation from the different domain distributions. We attempt to verify whether TBiGAN can learn a latent code space between domains by the objective we define in (3), (6) and (7).
For MNIST\(\rightarrow \)USPS, MNIST\(\rightarrow \)MNIST_m in Table 3, the generative model networks only contain several fully connected layers, the discriminator and Encoder both have the same structure with the generator. Since MNIST and USPS have similar domain distributions, a relatively simple network structure is proposed.
For MNIST\(\rightarrow \)MNIST_m in Table 4, we define a different network like conv-pool-conv-pool-fc-softmax. The Discriminator contains three conv-pool layers followed by two fully connected layers (depends on the different image preprocessing methods) activated by sigmoid units. In particular the Encoder for MNIST_m domain only has two hidden layers activated by ReLU units. A fully connected layer still be used as the last output layer.
Since SVHN has its own domain-specific properties, a single image contains several adjacent digits. The architectures of network need more convolutional layers to capture the domain information. Therefore, the discriminator has five conv-pool layers followed by last two full-connected layers activated with a sigmoid unit. The specific details of the Generator and Encoder are shown in Table 5.
1.2 A.2 Cycle-Consistent Crossing Domain Translation
The fundamental network architectures of Cycle-TBiGAN are similar to TBiGAN. We assume that necessary components such as generators (\(G_S,\ G_T\)) and encoder (E) corresponding to specific domain have been obtained from TBiGAN. The Generator \(G_c(\cdot )\) is actually a translator that maps the latent code space of target domain \(Z_T\) and source domain \(Z_S\) to a synthesized code space \(Z_{syn}\), which means the invariant feature representation space is regarded as input for \(G_c(\cdot )\). A specific network description of mapping function \(G_c(\cdot )\) is showed in Table 6.
As defined in (16), the \(L_m (\cdot )\) is a “similarity measurement”, which is used to find a subset of \(z_{s_j}\) that similar to target image latent code \(z_{t_i}\). Since the training of Cycle-TBiGAN is high computational cost and it should be relaxed, we use the K-Nearest-Neighbor (KNN) algorithm to find a subset of \(z_{s_{1\cdot \cdot \cdot k}}\) with size k from source latent code space. In other words, the latent subset should be similar to \(z_{t_i}\). Therefore, the relaxed (16) could be presented as:
The relaxed objective could be optimized using SGD.
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Wei, Y. (2019). Transferable Adversarial Cycle Alignment for Domain Adaption. In: Tetko, I., Kůrková, V., Karpov, P., Theis, F. (eds) Artificial Neural Networks and Machine Learning – ICANN 2019: Deep Learning. ICANN 2019. Lecture Notes in Computer Science(), vol 11728. Springer, Cham. https://doi.org/10.1007/978-3-030-30484-3_52
Download citation
DOI: https://doi.org/10.1007/978-3-030-30484-3_52
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-30483-6
Online ISBN: 978-3-030-30484-3
eBook Packages: Computer ScienceComputer Science (R0)