Communication-Compressed Adaptive Gradient Method for Distributed Nonconvex Optimization

Keyphrases

• Adaptive Gradient Methods 100%
• Communication Efficiency 66%
• Communication Cost 66%
• Distributed Learning 66%
• AMSGrad 66%
• First-order 33%
• Training Data 33%
• Stationary Point 33%
• Local Labour 33%
• Learning Framework 33%
• Vanilla 33%
• Central Server 33%
• Machine Learning Models 33%
• Model Update 33%
• Stochastic Gradient Descent 33%
• Iteration Complexity 33%
• Nonconvex Optimization Problems 33%
• Compression Strategy 33%
• Nonconvex Stochastic Optimization 33%
• Error Feedback 33%
• Optimization Setting 33%
• Large-scale Machine Learning 33%

Computer Science

• Gradient Method 100%
• Communication Cost 66%
• Distributed Learning 66%
• Optimization Problem 33%
• Training Dataset 33%
• Gradient Descent 33%
• Learning Framework 33%
• Stationary Point 33%