Fast Curvature Matrix-Vector Products for Second-Order Gradient Descent.
Nicol N. SchraudolphPublished in: Neural Comput. (2002)
Keyphrases
- sparse matrix
- gradient vector
- eigenvalues and eigenvectors
- higher order
- multiscale
- objective function
- loss function
- linearly independent
- rows and columns
- scale space
- cost function
- singular value decomposition
- weight vector
- feature vectors
- matrix representation
- symmetric matrices
- fuzzy membership
- curvature estimation
- dot product
- image processing
- weight matrix
- transformation matrix
- covariance matrix
- vector space
- high curvature
- product information
- linear algebra
- low rank
- neural network