Abstract
Learning results depend on the representation of data, so how to efficiently represent data has been a research hot spot in machine learning and artificial intelligence. With the deepening of the deep learning research, studying how to train the deep networks to express high dimensional data efficiently also has been a research frontier. In order to present data more efficiently and study how to express data through deep networks, we propose a novel stacked denoising sparse autoencoder in this paper. Firstly, we construct denoising sparse autoencoder through introducing both corrupting operation and sparsity constraint into traditional autoencoder. Then, we build stacked denoising sparse autoencoders which has multi-hidden layers by layer-wisely stacking denoising sparse autoencoders. Experiments are designed to explore the influences of corrupting operation and sparsity constraint on different datasets, using the networks with various depth and hidden units. The comparative experiments reveal that test accuracy of stacked denoising sparse autoencoder is much higher than other stacked models, no matter what dataset is used and how many layers the model has. We also find that the deeper the network is, the less activated neurons in every layer will have. More importantly, we find that the strengthening of sparsity constraint is to some extent equal to the increase in corrupted level.
Similar content being viewed by others
Notes
MNIST dataset http://yann.lecun.com/exdb/mnist/.
MNIST-Rotation dataset http://www.iro.umontreal.ca/~lisa/twiki/bin/view.cgi/Public/MnistVariations.
STL-10 dataset http://cs.stanford.edu/~acoates/stl10/.
M. Schmidt. minFunc: unconstrained differentiable multivariate optimization in MATLAB. http://www.cs.ubc.ca/~schmidtm/Software/minFunc.html, 2005 http://www.cs.ubc.ca/~schmidtm/Software/minFunc.html.
References
Bengio Y, Courville A, Vincent P (2013) Representation learning: a review and new perspectives. IEEE Trans Pattern Anal Mach Intell 35(8):1798–1828
Hinton GE, Osindero S, Teh YW (2006) A fast learning algorithm for deep belief nets. Neural Comput 18(7):1527–1554
Hinton GE (2007) Learning multiple layers of representation. Trends Cogn Sci 11(10):428–434
Bengio Y (2009) Learning deep architectures for AI. Found Trends Mach Learn 2(1):1–127
Hinton GE (2007) To recognize shapes, first learn to generate images. Prog Brain Res 165:535–547
Hinton GE, Srivastava N, Krizhevsky A, Sutskever I, Salakhutdinovet RR (2012) Improving neural networks by preventing co-adaptation of feature detectors. The computing research repository, 2012.1207(0580)
Vincent P, Larochelle H, Lajoie I, Bengio Y, Manzagol PA (2010) Stacked denoising autoencoders: learning useful representations in a deep network with a local denoising criterion. J Mach Learn Res 11:3371–3408
Vincent P, Larochelle H, Bengio Y, Manzagol PA (2008) Extracting and composing robust features with denoising autoencoders. ACM, New York, pp 1096–1103
Krizhevsky A, Hinton GE (2010) Using very deep autoencoders for content-based image retrieval. Belgium: i6doc.com publication, pp 489–494
Hinton GE, Krizhevsky A, Wang SD (2011) Transforming auto-encoders. Springer, Espoo, pp 44–51
Hinton GE, Salakhutdinov RR (2006) Reducing the dimensionality of data with neural networks. Science 313(5786):504–507
Baldi P (2012) Autoencoders, unsupervised learning, and deep architectures. IEEE Transactions on Neural Networks, Bellevue, pp 37–50
Zhang N, Ding S, Shi Z (2016) Denoising Laplacian multi-layer extreme learning machine. Neurocomputing. 171(C):1066–1074.
Zeiler MD, Krishnan D, Taylor GW, Fergus R (2010) Deconvolutional networks. IEEE Computer Society, San Francisco, pp 2528–2535
Zeiler MD, Taylor GW, Fergus R (2011) Adaptive deconvolutional networks for mid and high level feature learning. Piscataway, NJ, USA: IEEE. 2011.2018-25
Serre T, Wolf L, Bileschi S, Riesenhuber M, Poggio T (2007) Robust object recognition with cortex-like mechanisms. IEEE Trans Pattern Anal Mach Intell 29(3):411–426
LeCun Y, Bengio Y (1995) Convolutional networks for images, speech, and time series. The handbook of brain theory and neural networks, p 3361
Lecun Y, Bottou L, Bengio Y, Haffner P (1998) Gradient-based learning applied to document recognition. Proc IEEE 86(11):2278–2324
Jarrett K, Kavukcuoglu K, Ranzato MA, LeCun Y (2009) What is the best multi-stage architecture for object recognition?. IEEE, Kyoto, pp 2146–2153
Le QV, Ranzato AM, Monga R, Devin M, Chen K, Corrado GS, Dean J, Ng AY (2013) Building high level features using large scale unsupervised learning. IEEE, New York, pp 8595–8598
Chandra B, Sharma RK (2014) Adaptive noise schedule for denoising autoencoder. Springer International Publishing, Cham, pp 535–542
Memisevic R (2011) Gradient-based learning of higher-order image features. Barcelona, Spain: 2011 IEEE international conference on computer vision (ICCV), pp 1591–1598
Lemme A, Reinhart RF, Steil JJ (2012) Online learning and generalization of parts-based image representations. Neural Netw 33:194–203
Deng J, Zhang Z, Marchi E, Schuller B (2013) Sparse autoencoder-based feature transfer learning for speech emotion recognition. IEEE, New York, pp 511–516
Tan CC, Eswaran C (2011) Using autoencoders for mammogram compression. J Med Syst 35(1):49–58
Lee MK, Han DS (2012) Dimensionality reduction of radio map with nonlinear autoencoder. Electron Lett 48(11):655–657
Liu J, Chi G, Liu Z, Liu Y, Li H, Luo X (2013) Predicting protein structural classes with autoencoder neural networks. IEEE, Piscataway, pp 1894–1899
Gu B, Sheng VS, Tay KY, Romano W, Li S (2015) Incremental Support Vector Learningfor Ordinal Regression. IEEE Transactions on Neural Networks and Learning Systems, vol. 26, no. 7, pp. 1403–1416
Palecek K (2014) Extraction of features for lip-reading using autoencoders. Springer, Novi Sad, pp 209–216
Xia Z, Wang X, Sun X, Wang B (2014) Steganalysis of least significant bit matching using multi-order differences. Secur Commun Netw. 7(8):1283–1291
Jiang X, Zhang Y, Zhang W, Xiao X (2013) A novel sparse auto-encoder for deep unsupervised learning. In: 2013 sixth international conference on advanced computational intelligence (ICACI). IEEE Press, Hangzhou, pp 256–261
Le QV, Ngiam J, Coates A, Lahiri A, Prochnow B, Ng AY (2011) On optimization methods for deep learning. Proceedings of the 28th international conference on machine learning (ICML-11), pp 265–272
Xie J, Xu L, Chen E (2012) Image denoising and inpainting with deep neural networks. Nips, pp 341–349
Cho K (2013) Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. In: Proceedings of the 30th international conference on machine learning (ICML-13), pp 432–440
Chen Y, Hao C, Wu W, Wu E (2016) Robust dense reconstruction by range merging based on confidence estimation. Sci China Inf Sci. 59(9)
Li J-C, Ng WWY, Yeung DS, Chan PPK (2014) Bi-firing deep neural networks. Int J Mach Learn Cybernet 5(1):73–83
Coates A, Lee H, Ng AY (2011) An analysis of single layer networks in unsupervised feature learning. AISTATS, pp 215–223
Makhzani A, Frey B (2014) K-sparse autoencoders. International conference on learning representations, ICLR
Vagin V, Fomina M (2011) Problem of knowledge discovery in noisy databases. Int J Mach Learn Cybernet 2(3):135–145
Acknowledgements
The work is partially supported by the National Natural Science Foundation of China under Grant No. 61379101 and No.61672522, and the National Basic Research Program of China under Grant No. 2013CB329502.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Meng, L., Ding, S., Zhang, N. et al. Research of stacked denoising sparse autoencoder. Neural Comput & Applic 30, 2083–2100 (2018). https://doi.org/10.1007/s00521-016-2790-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-016-2790-x