Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition
| Autor: | Y-Lan Boureau, Marc'Aurelio Ranzato, Fu Jie Huang, Yann LeCun |
|---|---|
| Rok vydání: | 2007 |
| Předmět: |
Caltech 101
Computer science business.industry Feature extraction Supervised learning Cognitive neuroscience of visual object recognition Convolutional Deep Belief Networks Pattern recognition Sigmoid function Object detection Unsupervised learning Artificial intelligence Invariant (mathematics) business MNIST database |
| Zdroj: | CVPR |
| DOI: | 10.1109/cvpr.2007.383157 |
| Popis: | 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 with 30 training samples per category. While the resulting architecture is similar to convolutional networks, the layer-wise unsupervised training procedure alleviates the over-parameterization problems that plague purely supervised learning procedures, and yields good performance with very few labeled training samples. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|