A Coordinate-Descent Primal-Dual Algorithm with Large Step Size and Possibly Nonseparable Functions.
Olivier FercoqPascal BianchiPublished in: SIAM J. Optim. (2019)
Keyphrases
- step size
- stochastic gradient descent
- cost function
- convergence rate
- convergence speed
- evolutionary programming
- steepest descent method
- faster convergence
- line search
- logistic regression
- hessian matrix
- quantization step
- wavelet transform
- gradient method
- multiple kernel learning
- variable step size
- feature extraction
- high frequency
- high quality