Corpus ID: 14003079

A Winner-Take-All Method for Training Sparse Convolutional Autoencoders

@article{Makhzani2014AWM,
  title={A Winner-Take-All Method for Training Sparse Convolutional Autoencoders},
  author={Alireza Makhzani and B. Frey},
  journal={ArXiv},
  year={2014},
  volume={abs/1409.2752}
}
We explore combining the benefits of convolutional architec tures and autoencoders for learning deep representations in an unsupervise d manner. A major challenge is to achieve appropriate sparsity among hidden v ariables, since neighbouring variables in each feature map tend to be highly corre lated and a suppression mechanism is therefore needed. Previously, decon volutional networks and convolutional predictive sparse decomposition have be en used to construct systems that have a recognition… Expand
29 Citations

Paper Mentions

Conditional Generative Denoising Autoencoder
  • 1
  • PDF
Generative Moment Matching Networks
  • 526
  • PDF
Stacked What-Where Auto-encoders
  • 216
  • PDF
Pre-Training CNNs Using Convolutional Autoencoders
  • 7
  • PDF
EE-AE: An Exclusivity Enhanced Unsupervised Feature Learning Approach
  • Jingcai Guo, Song Guo
  • Computer Science, Mathematics
  • ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2019
  • 1
  • PDF
Learning Sparse, Distributed Representations using the Hebbian Principle
  • 5
  • PDF
The effect of whitening transformation on pooling operations in convolutional autoencoders
  • 16
Bottom-up Deep Learning using the Hebbian Principle Aseem
  • 5
  • PDF
Structured Sparse Convolutional Autoencoder
  • PDF
Shallow sparse autoencoders versus sparse coding algorithms for image compression
  • 1
  • PDF
...
1
2
3
...

References

SHOWING 1-10 OF 28 REFERENCES
k-Sparse Autoencoders
  • 241
  • PDF
An Analysis of Single-Layer Networks in Unsupervised Feature Learning
  • 2,083
  • Highly Influential
  • PDF
Selecting Receptive Fields in Deep Networks
  • 212
  • PDF
Convolutional Deep Belief Networks on CIFAR-10
  • 360
  • PDF
ImageNet classification with deep convolutional neural networks
  • 61,159
  • PDF
Learning Convolutional Feature Hierarchies for Visual Recognition
  • 528
  • Highly Influential
  • PDF
Differentiable Pooling for Hierarchical Feature Learning
  • 13
  • Highly Influential
  • PDF
Factored 3-Way Restricted Boltzmann Machines For Modeling Natural Images
  • 228
  • PDF
...
1
2
3
...