Regularized Auto-Encoders Estimate Local Statistics

What do auto-encoders learn about the underlying data generating distribution? Recent work suggests that some auto-encoder variants do a good job of capturing the local manifold structure of the unknown data generating density. This paper clarifies some of these previous intuitive observations by showing that minimizing a particular form of regularized reconstruction error yields a reconstruction function that locally characterizes the shape of the data generating density. More precisely, we show that the auto-encoder captures the score (derivative of the logdensity with respect to the input) or the local mean associated with the unknown data-generating density. This is the second result linking denoising auto-encoders and score matching, but in way that is different from previous work, and can be applied to the case when the auto-encoder reconstruction function does not necessarily correspond to the derivative of an energy function. The theorems provided here are completely generic and do not depend on the parametrization of the autoencoder: they show what the auto-encoder would tend to if given enough capacity and examples. These results are for a contractive training criterion we show to be similar to the denoising auto-encoder training criterion with small corruption noise, but with contraction applied on the whole reconstruction function rather than just encoder. Similarly to score matching, one can consider the proposed training criterion as a convenient alternative to maximum likelihood, i.e., one not involving a partition function. Finally, we make the connection to existing sampling algorithms for such autoencoders, based on an MCMC walking near the high-density manifold.

[1]  David J. Field,et al.  Sparse coding with an overcomplete basis set: A strategy employed by V1? , 1997, Vision Research.

[2]  Christian P. Robert,et al.  Monte Carlo Statistical Methods , 2005, Springer Texts in Statistics.

[3]  B. Dacorogna.  Introduction to the calculus of variations , 2004 .

[4]  Lawrence Cayton,et al.  Algorithms for manifold learning , 2005 .

[5]  Yee Whye Teh,et al.  A Fast Learning Algorithm for Deep Belief Nets , 2006, Neural Computation.

[6]  Marc'Aurelio Ranzato,et al.  Efficient Learning of Sparse Representations with an Energy-Based Model , 2006, NIPS.

[7]  Marc'Aurelio Ranzato,et al.  Sparse Feature Learning for Deep Belief Networks , 2007, NIPS.

[8]  Thomas Hofmann,et al.  Greedy Layer-Wise Training of Deep Networks , 2007 .

[9]  Yoshua. Bengio,et al.  Learning Deep Architectures for AI , 2007, Found. Trends Mach. Learn..

[10]  Yoshua Bengio,et al.  Extracting and composing robust features with denoising autoencoders , 2008, ICML '08.

[11]  H. Sebastian Seung,et al.  Natural Image Denoising with Convolutional Networks , 2008, NIPS.

[12]  Honglak Lee,et al.  Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations , 2009, ICML '09.

[13]  R. Fergus,et al.  Learning invariant features through topographic filter maps , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[14]  Geoffrey E. Hinton,et al.  Deep Boltzmann Machines , 2009, AISTATS.

[15]  A. Hyvärinen,et al.  Estimation of Non-normalized Statistical Models , 2009 .

[16]  Yann LeCun,et al.  Regularized estimation of image statistics by Score Matching , 2010, NIPS.

[17]  Hariharan Narayanan,et al.  Sample Complexity of Testing the Manifold Hypothesis , 2010, NIPS.

[18]  Yann LeCun,et al.  Structured sparse coding via lateral inhibition , 2011, NIPS.

[19]  Pascal Vincent,et al.  Contractive Auto-Encoders: Explicit Invariance During Feature Extraction , 2011, ICML.

[20]  Nando de Freitas,et al.  On Autoencoders and Score Matching for Energy Based Models , 2011, ICML.

[21]  Pascal Vincent,et al.  A Connection Between Score Matching and Denoising Autoencoders , 2011, Neural Computation.

[22]  Pascal Vincent,et al.  The Manifold Tangent Classifier , 2011, NIPS.

[23]  Yoshua Bengio,et al.  On the Expressive Power of Deep Architectures , 2011, ALT.

[24]  Yoshua Bengio,et al.  A Generative Process for sampling Contractive Auto-Encoders , 2012, ICML 2012.

[25]  Yoshua Bengio,et al.  Implicit Density Estimation by Local Moment Matching to Sample from Auto-Encoders , 2012, ArXiv.

[26]  Pascal Vincent,et al.  Unsupervised Feature Learning and Deep Learning: A Review and New Perspectives , 2012, ArXiv.