Projective Proximal Gradient Descent for Nonconvex Nonsmooth Optimization: Fast Convergence Without Kurdyka-Lojasiewicz (KL) Property.
Yingzhen YangPing LiPublished in: ICLR (2023)
Keyphrases
- globally convergent
- image restoration and reconstruction
- global convergence
- optimization problems
- variational inequalities
- global optimization
- line search
- nonlinear programming
- autocalibration
- cost function
- objective function
- newton method
- conjugate gradient
- faster convergence
- stationary points
- augmented lagrangian
- evolutionary multiobjective
- mathematical programming
- optimization methods
- markov random field
- global optimum
- convergence rate
- update rule
- optimization algorithm
- convergence speed
- constrained optimization
- convergence analysis
- image restoration
- mathematical program
- evolutionary algorithm