An Incremental Gradient(-Projection) Method with Momentum Term and Adaptive Stepsize Rule.
Paul TsengPublished in: SIAM J. Optim. (1998)
Keyphrases
- step size
- gradient projection method
- variable step size
- convergence rate
- faster convergence
- approximate dynamic programming
- learning rate
- cost function
- multiscale
- neural network
- convergence speed
- search direction
- objective function
- regularization term
- temporal difference
- incremental learning
- hessian matrix
- quasi newton