Convergence of Gradient Dynamics with a Variable Learning Rate.
Michael H. BowlingManuela M. VelosoPublished in: ICML (2001)
Keyphrases
- learning rate
- convergence rate
- rapid convergence
- convergence theorem
- convergence speed
- gradient method
- update rule
- adaptive learning rate
- weight update
- error function
- learning algorithm
- global convergence
- hidden layer
- initial conditions
- step size
- dynamical systems
- training algorithm
- multilayer neural networks
- uniform convergence
- bp neural network algorithm
- activation function
- weight vector
- line search
- primal dual
- differential evolution
- linear program
- particle swarm optimization
- high accuracy
- optimal solution
- genetic algorithm