ML Anthology

Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition

Marc'Aurelio Ranzato, Fu Jie Huang, Y-Lan Boureau, Yann LeCun
CVPR 2007
doi:10.1109/CVPR.2007.383157 /cvpr/2007/ranzato2007cvpr-unsupervised/

Abstract

We present an unsupervised method for learning a hierarchy of sparse feature detectors that are invariant to small shifts and distortions. The resulting feature extractor consists of multiple convolution filters, followed by a feature-pooling layer that computes the max of each filter output within adjacent windows, and a point-wise sigmoid non-linearity. A second level of larger and more invariant features is obtained by training the same algorithm on patches of features from the first level. Training a supervised classifier on these features yields 0.64 % error on MNIST, and 54 % average recognition rate on Caltech 101

PDF CVPR Semantic Scholar

Cite

Text

Ranzato et al. "Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2007. doi:10.1109/CVPR.2007.383157

Markdown

[Ranzato et al. "Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2007.](https://mlanthology.org/cvpr/2007/ranzato2007cvpr-unsupervised/) doi:10.1109/CVPR.2007.383157

BibTeX

@inproceedings{ranzato2007cvpr-unsupervised,
  title     = {{Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition}},
  author    = {Ranzato, Marc'Aurelio and Huang, Fu Jie and Boureau, Y-Lan and LeCun, Yann},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2007},
  doi       = {10.1109/CVPR.2007.383157},
  url       = {https://mlanthology.org/cvpr/2007/ranzato2007cvpr-unsupervised/}
}