Linearized ADMM Converges to Second-Order Stationary Points for Non-Convex Problems

Mathematics

• Stationary Point 100%
• Nonconvex Problem 100%
• Loss Function 50%
• Probability Theory 50%
• Neural Network 50%
• Matrix (Mathematics) 50%
• Signal Processing 50%
• Factorization 50%
• Dual Variable 50%
• Lipschitz Condition 50%

Keyphrases

• Linearized ADMM 100%
• Primal-dual Method of multipliers 50%
• Decentralized Matrix Factorization 50%

Computer Science

• Lipschitz Condition 50%