Global convergence and variance reduction for a class of nonconvex-nonconcave minimax problems

Keyphrases

• Adversarial Learning 20%
• Alternating Gradient Descent 60%
• Ascent Algorithm 40%
• Constant Step Size 20%
• Convex-concave 20%
• Fast Rates 20%
• Finite Sum 20%
• Generative Adversarial Networks 20%
• Global Convergence 100%
• Global Variance 100%
• Gradient Ascent 100%
• Linear Growth Rate 20%
• Machine Learning Applications 20%
• Minimax Problem 100%
• Networked Learning 20%
• Nonconcave 100%
• Nonconvex 100%
• Nonconvex Games 20%
• Ojasiewicz Inequality 20%
• SIMPLE Algorithm 20%
• Vanilla Gradient Descent 20%
• Variance Reduction 100%

Computer Science

• Algorithm Converges 20%
• Common Practice 20%
• Generative Adversarial Networks 20%
• Global Convergence 100%
• Gradient Descent 100%
• Machine Learning 20%
• Minimax Problem 100%
• Reduced Algorithm 20%
• Simple Algorithm 20%
• Subclasses 20%
• Variance Reduction 100%

Mathematics

• Finite Sum 50%
• Minimax 100%
• Stochastics 50%
• Variance 50%
• Variance Reduction 100%