Training Neural Networks with an Extension of the Marquardt Algorithm.
Markus LendlRolf UnbehauenPublished in: NC (1998)
Keyphrases
- preprocessing
- experimental evaluation
- dynamic programming
- neural network
- recognition algorithm
- optimization algorithm
- detection algorithm
- probabilistic model
- training algorithm
- theoretical analysis
- neural network training
- training process
- high accuracy
- worst case
- computational cost
- artificial neural networks
- np hard
- computational complexity
- similarity measure
- cost function
- significant improvement
- search space
- simulated annealing
- matching algorithm
- objective function
- improved algorithm
- data sets
- backpropagation algorithm
- learning algorithm
- times faster
- self organizing maps
- optimal solution
- training set