Newton-Based Trainable Learning Rate.
George RetsinasGiorgos SfikasPanagiotis Paraskevas FilntisisPetros MaragosPublished in: ICASSP (2023)
Keyphrases
- learning rate
- convergence rate
- learning algorithm
- hidden layer
- error function
- rapid convergence
- convergence speed
- adaptive learning rate
- least squares
- interior point methods
- multilayer neural networks
- training algorithm
- activation function
- primal dual
- weight vector
- newton method
- delta bar delta
- linear combination
- convergence theorem
- neural network
- state space
- evolutionary algorithm
- pairwise
- data mining