Analysis of parallel training algorithms for deep neural networks

Mathematics

• Training Algorithm 100%
• Stochastic Gradient 80%
• Parallel Algorithms 68%
• Gradient Descent 60%
• Neural Networks 57%
• Descent Algorithm 31%
• Gradient Algorithm 29%
• Gradient Descent Method 17%
• Parallelization 15%
• Training 12%
• Speedup 12%
• Approximation 6%

Business & Economics

• Neural Networks 73%
• Gradient 67%
• Train 11%
• Approximation 10%

Engineering & Materials Science

• Deep neural networks 59%
• Steepest descent method 19%