Because of the large structure and long training time, the development cycle of the common depth model is prolonged. How to speed up training is a problem deserving of study. In order to accelerate training, K-means clustering optimizing deep stacked sparse autoencoder (K-means sparse SAE) is presented in this paper. First, the input features are divided into K small subsets by K-means clustering, then each subset is input into corresponding autoencoder model for training, which only has fewer nodes in the hidden layer than traditional models. After training, each autoencoder’s trained weights and biases is merged to obtain the next layer’s input features by feedforward network. The above steps are repeated till the softmax layer, then fine-tuning is carried out. Using MNIST-Rotation datasets to train the network that has three hidden layers and each layer has 800 nodes, the improved model has higher classification accuracy and shorter training time when K = 10. With K increasing, the training time is reduced to almost the same as the fine-tuning time but the recognition ability is descended. Compared with the recently stacked denoising sparse autoencoder, the recognition accuracy is improved by 1%, not only the noise factor is not selected but also the training speed is significantly increased. The trained filters from the improved model is also used to train convolutional autoencoder, and it performs better than traditional models. We find that pre-training stage doesn’t need large samples simultaneously, and small samples parallel training reduces the probability of falling into the local minimum.
This is a preview of subscription content, log in to check access.
Buy single article
Instant access to the full article PDF.
Price includes VAT for USA
Subscribe to journal
Immediate online access to all issues from 2019. Subscription will auto renew annually.
This is the net price. Taxes to be calculated in checkout.
MNIST dataset http://yann.lecun.com/exdb/mnist/.
MNIST-Rotation dataset http://www.iro.umontreal.ca/%7elisa/twiki/bin/view.cgi/Public/MnistVariations.
The reduced STL-10 dataset http://ufldl.stanford.edu/wiki/resources/.
M. Schmidt. minFunc: unconstrained differentiable multivariate optimization in MATLAB. http://www.cs.ubc.ca/%7eschmidtm/Software/minFunc.html, 2005 http://www.cs.ubc.ca/%7eschmidtm/Software/minFunc.html.
Alain, G., & Bengio, Y. (2012). What regularized auto-encoders learn from the data generating distribution. Computer Science, 15(1), 3563–3593.
Bell, A. J. (1996). Edges are the ’independent components’ of natural scenes. In: Advances in neural information processing system (pp. 831–837).
Bellinger, C., Drummond, C., & Japkowicz, N. (2017). Manifold-based synthetic oversampling with manifold conformance estimation. Machine Learning, 1, 1–33.
Chandra, B., & Sharma, R. K. (2014). Adaptive noise schedule for denoising autoencoder. In: International conference on neural information processing (pp. 535–542).
Cheng, X., Liu, H., Xu, X., & Sun, F. (2016). Denoising deep extreme learning machine for sparse representation. Memetic Computing, 9(3), 1–14.
Das, R., & Walia, E. (2017). Partition selection with sparse autoencoders for content based image classification. Neural Computing & Applications, 4, 1–16.
Glorot, X., Bordes, A., & Bengio, Y. (2012). Deep sparse rectifier neural networks. Jmlr W & Cp, 15, 315–323.
Gupta, K., & Majumdar, A. (2017). Imposing class-wise feature similarity in stacked autoencoders by nuclear norm regularization. Neural Processing Letters, 2, 1–15.
Hinton, G. E., Osindero, S., & Teh, Y. W. (2006). A fast learning algorithm for deep belief nets. Neural Computation, 18(7), 1527–1554. https://doi.org/10.1162/neco.2006.18.7.1527.
Hong, C., Yu, J., You, J., Yu, Z., & Chen, X. (2017). Three-dimensional image-based human pose recovery with hypergraph regularized autoencoders. Multimedia Tools & Applications, 2016(1), 1–19.
Imakura, A., Inoue, Y., Sakurai, T., & Futamura, Y. (2018). Parallel implementation of the nonlinear semi-NMF based alternating optimization method for deep neural networks. Neural Processing Letters, 47(3), 815–827.
Le, Q. V. (2013). Building high-level features using large scale unsupervised learning. In: IEEE international conference on acoustics, speech and signal processing (pp. 8595–8598).
Lemme, A., Reinhart, R. F., & Steil, J. J. (2010). Efficient online learning of a non-negative sparse autoencoder. In: Esann 2010, European symposium on artificial neural networks, Bruges, Belgium, April 28–30, 2010 Proceedings.
Li, B., & Chen, C. (2017). First-order sensitivity analysis for hidden neuron selection in layer-wise training of networks. Neural Processing Letters, 7, 1–17.
Li, R., & Xu, H. (2017). Parallel stacked autoencoder and its application in process modeling. Journal of Electronic Measurement & Instrumentation, 31, 264–271.
Li, Z., Fan, Y., & Liu, W. (2015). The effect of whitening transformation on pooling operations in convolutional autoencoders. Eurasip Journal on Advances in Signal Processing, 2015(1), 37.
Makhzani, A., & Frey, B. (2014). k-sparse autoencoders. In ICLR. https://arxiv.org/pdf/1312.5663.pdf.
Meng, L., Ding, S., Zhang, N., & Zhang, J. (2018). Research of stacked denoising sparse autoencoder. Neural Computing & Applications, 30(7), 2083–2100.
Meng, Q., Catchpoole, D., Skillicom, D., & Kennedy, P. J. (2017). Relational autoencoder for feature extraction. In: International joint conference on neural networks (pp. 364–371).
Rifai, S., Vincent, P., Muller, X., Glorot, X., & Bengio, Y. (2011). Contractive auto-encoders: Explicit invariance during feature extraction. In: ICML.
Schlkopf, B., Platt, J., & Hofmann, T. (2006). Efficient learning of sparse representations with an energy-based model. In: Advances in neural information processing systems (pp. 1137–1144).
Schlkopf, B., Platt, J., & Hofmann, T. (2006). Greedy layer-wise training of deep networks. In: International conference on neural information processing systems (pp. 153–160).
Shu, Z., Wu, X.J., & Hu, C. (2018). Structure preserving sparse coding for data representation. Neural Processing Letters, 48(3), 1705–1719.
Singhal, V., & Majumdar, A. (2017). Majorization minimization technique for optimally solving deep dictionary learning. Neural Processing Letters, 3, 1–16.
Zeng, N., Zhang, H., Song, B., Liu, W., Li, Y., & Dobaie, A. M. (2017). Facial expression recognition via learning deep sparse autoencoders. Neurocomputing, 273, 643–649.
The work is supported by National Key Technology Research and Development Program of China No.2011BAD21B0601
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Bi, Y., Wang, P., Guo, X. et al. K-Means Clustering Optimizing Deep Stacked Sparse Autoencoder. Sens Imaging 20, 6 (2019). https://doi.org/10.1007/s11220-019-0227-1
- K-means clustering
- Sparse autoencoder
- Convolutional autoencoder
- Training method