Maximal Initial Learning Rates in Deep ReLU Networks

Keyphrases

• Learning Rate 100%
• Deep ReLU Networks 100%
• Neural Network 33%
• Further Analysis 16%
• Input Layer 16%
• Constant Width 16%
• Fully Connected 16%
• ReLU Networks 16%
• Large Learning Rate 16%

Computer Science

• Learning Rate 100%
• Neural Network 28%

Chemical Engineering

• Neural Network 100%