KKT condition-based smoothing recurrent neural network for nonsmooth nonconvex optimization in compressed sensing.
Dan WangZhuhong ZhangPublished in: Neural Comput. Appl. (2019)
Keyphrases
- recurrent neural networks
- compressed sensing
- image restoration and reconstruction
- image reconstruction
- globally convergent
- optimization problems
- random projections
- complex valued
- neural network
- sparse representation
- nonlinear programming
- recurrent networks
- feed forward
- artificial neural networks
- natural images
- orthogonal matching pursuit
- variational inequalities
- convex optimization
- optimality conditions
- convex functions
- mathematical programming
- markov random field
- pattern recognition
- objective function
- signal processing
- signal recovery
- data sets