发布日期:May 30, 2014 6:29:42 AM
Abstract:
In this article, Deep Sparse Autoencoder(DSAE) which is one of the techniques of deep learning is evaluated as a feature extractor. To evaluate the DSAE,the extracted features obtained by the proposal technique and by Scale-invariant feature transform(SIFT) were classified using a latent dirichlet allocation(LDA). The performance of the feature extractor was compared from the view point of classification performance. As a result, DSAE classified an object individually.