File Exchange

image thumbnail

Sparse Autoencoder

version 1.2.0 (36.5 KB) by BERGHOUT Tarek
These codes returns a fully traned Sparse Autoencoder


Updated 10 Jul 2019

View License

I dedicate this work to my son ' BERGHOUT Loukmane '.
Sparse_AE: the function allows to train an Auto encoder In Sparse domain by solving L1 norm optimization problem. Optimization problem: min_B ||B||_1 subject to : H*B = X.
Where H: is the Sparse version hidden layer, B: is output weights matrix and X:is the Input.
the white gaussian noise generated from a random distribution must be added during training , To make the sparse hidden layer a full rank matrix otherwise, L1 norm optimization will never achieved (check the Sparse_AE function .
for any references you can use these ones:
[1] R. G. Baraniuk, “<Compressive Sensing(lecture notes).pdf>,” no. July, pp. 118–121, 2007.
[2] M. W. Fakhr, E. N. S. Youssef, and M. S. El-Mahallawy, “L1-regularized least squares sparse extreme learning machine for classification,” 2015 Int. Conf. Inf. Commun. Technol. Res. ICTRC 2015, no. April, pp. 222–225, 2015.
[3] G. Huang, S. Member, H. Zhou, X. Ding, and R. Zhang, “Extreme Learning Machine for Regression and Multiclass Classification,” vol. 42, no. 2, pp. 513–529, 2012.
[4] C. justin Romberg and, “L1 magic toolbox.”
[5] A. Makhzani and B. Frey, “k-Sparse Autoencoders,” 2013.
[6] L. le Cao, W. bing Huang, and F. chun Sun, “Building feature space of extreme learning machine with sparse denoising stacked-autoencoder,” Neurocomputing, vol. 174, pp. 60–71, 2016.
you can also learn from this video tutorial:

Cite As

BERGHOUT Tarek (2019). Sparse Autoencoder (, MATLAB Central File Exchange. Retrieved .

Comments and Ratings (2)

ok thanks for the advice

great job , I think that you need to add one more thing just to follow the norms,it is not that much of importance but you need to add it,the bias matrix



some details


new referances.

MATLAB Release Compatibility
Created with R2019a
Compatible with any release
Platform Compatibility
Windows macOS Linux