Assignment 2 - Deep Learning with Sparse Coding
暂无分享,去创建一个
In sparse coding and sparse modeling, the n-diensional input x is assumed to be linearly reconstructable by computing the product of a sparse m-dimensional vector z with an n ×m dictionary matrix W , whose columns are sometimes called basis functions, basis vectors, or atoms. The sparse code z for a given input vector x is often obtained by minimizing the energy function: E(W,x, z) = ‖x−Wz‖2 + λ‖z‖1,
[1] Geoffrey E. Hinton,et al. Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.
[2] Marc Teboulle,et al. A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems , 2009, SIAM J. Imaging Sci..
[3] Yoshua Bengio,et al. Deep Sparse Rectifier Neural Networks , 2011, AISTATS.
[4] Marc'Aurelio Ranzato,et al. Fast Inference in Sparse Coding Algorithms with Applications to Object Recognition , 2010, ArXiv.