A corrected Levenberg-Marquardt algorithm with a nonmonotone line search for the system of nonlinear equations.
Yedan HeChangfeng MaBin FanPublished in: Appl. Math. Comput. (2015)
Keyphrases
- levenberg marquardt
- trust region
- nonlinear equations
- learning algorithm
- line search
- convergence rate
- artificial neural networks
- conjugate gradient
- least squares
- dynamic programming
- neural network
- gauss newton
- optimization algorithm
- np hard
- cost function
- objective function
- back propagation
- bundle adjustment
- training algorithm
- maximum likelihood estimation
- fuzzy logic
- global convergence
- search space
- computer vision