Acceleration via Fractal Learning Rate Schedules.
Naman AgarwalSurbhi GoelCyril ZhangPublished in: CoRR (2021)
Keyphrases
- learning rate
- fractal dimension
- convergence rate
- scheduling problem
- learning algorithm
- image compression
- hidden layer
- error function
- rapid convergence
- activation function
- adaptive learning rate
- weight vector
- convergence speed
- training algorithm
- multilayer neural networks
- convergence theorem
- bp neural network algorithm
- delta bar delta