Abstract
Transfer learning, which focuses on finding a favorable representation for instances of different domains based on auxiliary data, can mitigate the divergence between domains through knowledge transfer. Recently, increasing efforts on transfer learning have employed deep neural networks (DNN) to learn more robust and higher level feature representations to better tackle cross-media disparities. However, only a few articles consider the correction and semantic matching between multi-layer heterogeneous domain networks. In this article, we propose a deep semantic mapping model for heterogeneous multimedia transfer learning (DHTL) using co-occurrence data. More specifically, we integrate the DNN with canonical correlation analysis (CCA) to derive a deep correlation subspace as the joint semantic representation for associating data across different domains. In the proposed DHTL, a multi-layer correlation matching network across domains is constructed, in which the CCA is combined to bridge each pair of domain-specific hidden layers. To train the network, a joint objective function is defined and the optimization processes are presented. When the deep semantic representation is achieved, the shared features of the source domain are transferred for task learning in the target domain. Extensive experiments for three multimedia recognition applications demonstrate that the proposed DHTL can effectively find deep semantic representations for heterogeneous domains, and it is superior to the several existing state-of-the-art methods for deep transfer learning.
- G. Andrew, R. Arora, J. Bilmes, and K. Livescu. 2013. Deep canonical correlation analysis. In International Conference on Machine Learning. 1247--1255. Google ScholarDigital Library
- M. Chen, Z. Xu, K. Weinberger, and F. Sha. 2012. Marginalized denoising autoencoders for domain adaptation. In International Conference on Machine Learning. 1--8. Google ScholarDigital Library
- Z. Ding, N. M. Nasrabadi, and Y. Fu. 2016. Task-driven deep transfer learning for image classification. In IEEE International Conference on Acoustics, Speech and Signal Processing. 2414--2418.Google Scholar
- J. Donahue, J. Hoffman, E. Rodner, K. Saenko, and T. Darrell. 2013. Semi-supervised domain adaptation with instance constraints. In IEEE Conference on Computer Vision and Pattern Recognition. 668--675. Google ScholarDigital Library
- L. Duan, D. Xu, and I. Tsang. 2012. Learning with augmented features for heterogeneous domain adaptation. In International Conference on Machine Learning. 711--718. Google ScholarDigital Library
- F. Feng, R. Li, and X. Wang. 2015. Deep correspondence restricted Boltzmann machine for cross-modal retrieval. Neurocomputing 154, 4 (2015), 50--60. Google ScholarDigital Library
- X. Glorot, A. Bordes, and Y. Bengio. 2011. Domain adaptation for large-scale sentiment classification: A deep learning approach. In International Conference on Machine Learning. 513--520. Google ScholarDigital Library
- D. R. Hardoon, S. Szedmak, and J. Shawe-Taylor. 2004. Canonical correlation analysis: An overview with application to learning methods. Neural Computation 16, 12 (2004), 2639--2664. Google ScholarDigital Library
- J. Hoffman, D. Wang, F. Yu, and T. Darrell. 2016. FCNS in the wild: Pixel-level adversarial and constraint-based adaptation. In Computer Vision and Pattern Recognition. arXiv preprint arXiv:1612.02649.Google Scholar
- L. Jing, C. Zhang, and M. K. Ng. 2012. SNMFCA: Supervised NMF-based image classification and annotation. IEEE Transactions on Image Processing 21, 11 (2012), 4508--4521. Google ScholarDigital Library
- B. Kulis, K. Saenko, and T. Darrell. 2011. What you saw is not what you get: Domain adaptation using asymmetric kernel transforms. In Computer Vision and Pattern Recognition. 1785--1792. Google ScholarDigital Library
- W. Li, L. Duan, D. Xu, and I. W. Tsang. 2014. Learning with augmented features for supervised and semi-supervised heterogeneous domain adaptation. IEEE Transactions on Pattern Analysis and Machine Intelligence 36, 6 (2014), 1134--1148. Google ScholarDigital Library
- M. Long, J. Wang, G. Ding, J. Sun, and P. S. Yu. 2014. Transfer joint matching for unsupervised domain adaptation. In IEEE Conference on Computer Vision and Pattern Recognition. 1410--1417. Google ScholarDigital Library
- J. Ngiam, A. Khosla, M. Kim, J. Nam, H. Lee, and A. Y. Ng. 2012. Multimodal deep learning. In International Conference on Machine Learning. 689--696. Google ScholarDigital Library
- J. Ni, Q. Qiu, and R. Chellappa. 2013. Subspace interpolation via dictionary learning for unsupervised domain adaptation. In IEEE Conference on Computer Vision and Pattern Recognition. 692--699. Google ScholarDigital Library
- S. J. Pan, I. W. Tsang, J. T. Kwok, and Q. Yang. 2011. Domain adaptation via transfer component analysis. IEEE Transactions on Neural Networks 22, 2 (2011), 199--210. Google ScholarDigital Library
- S. J. Pan and Q. Yang. 2010. A survey on transfer learning. IEEE Transactions on Knowledge 8 Data Engineering 22, 10 (2010), 1345--1359. Google ScholarDigital Library
- W. Pan and Q. Yang. 2013. Transfer learning in heterogeneous collaborative filtering domains. Artificial Intelligence 197, 4 (2013), 39--55. Google ScholarDigital Library
- J. C. Pereira, E. Coviello, G. Doyle, N. Rasiwasia, G. R. G. Lanckriet, R. Levy, and N. Vasconcelos. 2014. On the role of correlation and abstraction in cross-modal multimedia retrieval. IEEE Transactions Pattern Analysis and Machine Intelligence 36, 3 (2014), 521--535. Google ScholarDigital Library
- R. Raina, A. Battle, H. Lee, B. Packer, and A. Y. Ng. 2007. Self-taught learning: Transfer learning from unlabeled data. In Proceedings of the 24th International Conference on Machine Learning. 759--766. Google ScholarDigital Library
- H. Sagha, J. Deng, M. Gavryukova, J. Han, and B. Schuller. 2016. Cross lingual speech emotion recognition using canonical correlation analysis on principal component subspace. In IEEE International Conference on Acoustics, Speech and Signal Processing. 5800--5804.Google Scholar
- X. Shu, G. J. Qi, J. Tang, and J. Wang. 2015. Weakly-shared deep transfer networks for heterogeneous-domain knowledge propagation. In ACM International Conference on Multimedia. 35--44. Google ScholarDigital Library
- R. Socher, M. Ganjoo, H. Sridhar, O. Bastani, C. D. Manning, and A. Y. Ng. 2013. Zero-shot learning through cross-modal transfer. In Advances in Neural Information Processing Systems. 935--943. Google ScholarDigital Library
- J. Tang, X. Shu, Z. Li, G. J. Qi, and J. Wang. 2016. Generalized deep transfer networks for knowledge propagation in heterogeneous domains. ACM Transactions on Multimedia Computing Communications and Applications 12, 4s (2016), 68:1--68:22. Google ScholarDigital Library
- D. Tao, C. Hong, J. Yu, J. Wan, and M. Wang. 2015. Multimodal deep autoencoder for human pose recovery. IEEE Transactions on Image Processing 24, 12 (2015), 5659--5670.Google ScholarDigital Library
- E. Tzeng, J. Hoffman, K. Saenko, and T. Darrell. 2017. Adversarial discriminative domain adaptation. In Computer Vision and Pattern Recognition. 7167--7176.Google Scholar
- L. Yang, L. Jing, and M. K. Ng. 2015. Robust and non-negative collective matrix factorization for text-to-image transfer learning. IEEE Transactions on Image Processing 24, 12 (2015), 4701--4714.Google ScholarDigital Library
- L. Yang, L. Jing, J. Yu, and M. K. Ng. 2016. Learning transferred weights from co-occurrence data for heterogeneous transfer learning. IEEE Transactions on Neural Networks and Learning Systems 27, 11 (2016), 2187--2200.Google ScholarCross Ref
- Q. Yang, Y. Chen, G. R. Xue, W. Dai, and Y. Yu. 2009. Heterogeneous transfer learning for image clustering via the social web. In Joint Conference of the Meeting of the ACL and the International Joint Conference on Natural Language Processing. 1--9. Google ScholarDigital Library
- X. Yang, T. Zhang, C. Xu, and M. H. Yang. 2015. Boosted multifeature learning for cross-domain transfer. ACM Transactions on Multimedia Computing Communications and Applications 11, 3 (2015), 35:1--35:19. Google ScholarDigital Library
- Y. R. Yeh, C. H. Huang, and Y. C. Wang. 2014. Heterogeneous domain adaptation and classification by exploiting the correlation subspace. IEEE Transactions on Image Processing 23, 5 (2014), 2009--2018.Google ScholarCross Ref
- Q. Zhang, L. T. Yang, and Z. Chen. 2016. Deep computation model for unsupervised feature learning on big data. IEEE Transactions on Services Computing 9, 1 (2016), 161--171.Google Scholar
- Q. Zhang, H. Zhong, L. T. Yang, Z. Chen, and F. Bu. 2016. PPHOCFS: Privacy preserving high-order CFS algorithm on the cloud for clustering multimedia data. ACM Transactions on Multimedia Computing Communications and Applications 12, 4s (2016), 66:1--66:15. Google ScholarDigital Library
- X. Zhang, F. X. Yu, S. F. Chang, and S. Wang. 2015. Deep transfer network: Unsupervised domain adaptation. In Computer Vision and Pattern Recognition.Google Scholar
- L. Zhao, Z. Chen, and Z. J. Wang. 2018. Unsupervised multiview nonnegative correlated feature learning for data clustering. IEEE Signal Processing Letters 25, 1 (2018), 60--64.Google ScholarCross Ref
- L. Zhao, Z. Chen, Z. Yang, Y. Hu, and M. S. Obaidat. 2018. Local similarity imputation based on fast clustering for incomplete data in cyber-physical systems. IEEE Systems Journal 12, 2 (2018), 1610--1620.Google ScholarCross Ref
- J. T. Zhou, S. J. Pan, I. W. Tsang, and Y. Yan. 2014. Hybrid heterogeneous transfer learning through deep learning. In 28th AAAI Conference on Artificial Intelligence. 2213--2219. Google ScholarDigital Library
- J. T. Zhou, I. W. Tsang, S. J. Pan, and M. Tan. 2014. Heterogeneous domain adaptation for multiple classes. In Artificial Intelligence and Statistics. 1095--1103.Google Scholar
- Y. Zhu, Y. Chen, Z. Lu, S. J. Pan, G. R. Xue, Y. Yu, and Q. Yang. 2011. Heterogeneous transfer learning for image classification. In AAAI Conference on Artificial Intelligence. 1304--1309. Google ScholarDigital Library
- F. Zhuang, X. Cheng, P. Luo, S. J. Pan, and Q. He. 2015. Supervised representation learning: Transfer learning with deep autoencoders. In International Conference on Artificial Intelligence. 4119--4125. Google ScholarDigital Library
Index Terms
- Deep Semantic Mapping for Heterogeneous Multimedia Transfer Learning Using Co-Occurrence Data
Recommendations
Deep representation-based transfer learning for deep neural networks
AbstractIn recent years, deep neural networks (DNNs) have become the de facto models for practically all visual tasks and most temporal analysis tasks due to the abundance of available labeled data and advances in computational resources. Deep ...
Highlights- A deep representation-based transfer learning method is proposed for knowledge transfer between deep neural networks.
Analysis of transfer learning for deep neural network based plant classification models
Graphical abstractDisplay Omitted
Highlights- We examined the benefit of the transfer learning on plant classification.
- Four ...
AbstractPlant species classification is crucial for biodiversity protection and conservation. Manual classification is time-consuming, expensive, and requires experienced experts who are often limited available. To cope with these issues, ...
Transfer learning to detect neonatal seizure from electroencephalography signals
AbstractThis paper offers a solution to the problem of detecting neonatal seizures via a transfer learning technique that judiciously reconstructs pre-trained deep convolution neural networks (p-DCNN), including alexnet, resnet18, googlenet, densenet, and ...
Comments