Extracting and Composing Robust Features with Denoising Autoencoders
Abstract
Previous work has shown that the difficulties in learning deep generative or discriminative models can be overcome by an initial unsupervised learning step that maps inputs to useful itermediate representations. We introduce and motivate a new training principle for unsupervised learning of a representation based on the idea of making the learned representations robust to partial corruption of the input pattern. This approach can be used to train autoencoders, and these denoising autoencoders can be stacked to initialize deep architectures. The algorithm can be motivated from a manifold learning and information theoretic perspective or from a generative model perspective. Comparative experiments clearly show the surprising advantage of corrupting the input of autoencoders on a pattern classification benchmark suite.
Cite
Text
Vincent et al. "Extracting and Composing Robust Features with Denoising Autoencoders." International Conference on Machine Learning, 2008. doi:10.1145/1390156.1390294Markdown
[Vincent et al. "Extracting and Composing Robust Features with Denoising Autoencoders." International Conference on Machine Learning, 2008.](https://mlanthology.org/icml/2008/vincent2008icml-extracting/) doi:10.1145/1390156.1390294BibTeX
@inproceedings{vincent2008icml-extracting,
title = {{Extracting and Composing Robust Features with Denoising Autoencoders}},
author = {Vincent, Pascal and Larochelle, Hugo and Bengio, Yoshua and Manzagol, Pierre-Antoine},
booktitle = {International Conference on Machine Learning},
year = {2008},
pages = {1096-1103},
doi = {10.1145/1390156.1390294},
url = {https://mlanthology.org/icml/2008/vincent2008icml-extracting/}
}