Continuous vs. Discrete Optimization of Deep Neural Networks

Keyphrases

• Deep Neural Network 100%
• Discrete Optimization 100%
• Gradient Flow 100%
• Gradient Descent 75%
• Deep Learning 37%
• Flow Trajectory 25%
• Almost Surely 12%
• Global Minimum 12%
• Approximate Solution 12%
• Initial Value Problem 12%
• Computational Efficiency 12%
• Optimization Analysis 12%
• Degree of Approximation 12%
• Heterogeneous Activation 12%
• Random Initialization 12%
• Deep Linear Networks 12%

Mathematics

• Gradient Flow 100%
• Deep Neural Network 100%
• Deep Learning Method 37%
• Approximates 12%
• Numerical Solution 12%
• Neural Network 12%
• Initial-Value Problem 12%
• Open Question 12%
• Degree of Approximation 12%
• Step Size 12%

Chemical Engineering

• Deep Neural Network 100%
• Deep Learning Method 100%
• Neural Network 33%