The effect of Leaky ReLUs on the training and generalization of overparameterized networks

Keyphrases

• Overparameterized Network 100%
• Leaky ReLU 100%
• Training Error 100%
• Generalization Error 66%
• Leaky Rectified Linear Unit 66%
• Neural Network 33%
• Numerical Experiments 33%
• Activation Function 33%
• Absolute Value 33%
• Convergence Rate 33%
• Error Bound 33%
• Generalization Error Bound 33%
• Value Activation 33%
• Overparameterized Neural Networks 33%

Mathematics

• Error Bound 100%
• Neural Network 100%
• Upper Bound 50%
• Wide Class 50%
• Absolute Value 50%
• Numerical Experiment 50%
• Rate of Convergence 50%