Loading...

Automatically Learning of Image Features by Using Deep Sparse Networks

Shahin Shamsabadi, Ali | 2016

964 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 49085 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Babaie-Zadeh, Massoud; Rabiee, Hamid Reza
  7. Abstract:
  8. Data representation plays an important role in machine learning and the performance of machine learning algorithms for instance, in supervised learnings (e.g. classifcation), and unsupervised ones (e.g. image denoising), are heavily influenced by the input applied to them. Regarding the fact that data usually lacks the desirable quality, efforts are always made to make a more desirable representation of data to be used as input to machine learning algorithms. Among many different representation of data, sparse data representation preserves much more information about data while it is simpler than data. We proposed a new stacked sparse autoencoder by imposing power two of smooth L0 norm of data representation on the hidden layer of regular autoencoder. Smooth L0 squared not only increases the sparsity of data representation but also guarantees the sparsity of each data representation. We also proposed stacked sparse autoencoder to learn representations for image classifcation and denoising. When applied to the MNIST, CIFAR-10 and OPTDIGITS datasets, we fnd that this model guarantees that each data input has a sparse representation (tackle the limitation of sparse autoencoder) that leads to achieves better classifcation results and increase the ability of denoising images. Using the proposed sparse autoencoder, once the model parameters are learned, the sparse representation of any new data is obtained simply by a matrixvector multiplication without doing any optimiation which is necessary in sparse coding algorithm
  9. Keywords:
  10. Neural Network ; Representation Learning ; Sparse Representation ; Deep Learning ; Deep Networks ; Autoencoder

 Digital Object List

 Bookmark

...see more