Convergence Analyses on On-Line Weight Noise Injection-Based Training Algorithms for MLPs.
John SumChi-Sing LeungKevin HoPublished in: IEEE Trans. Neural Networks Learn. Syst. (2012)
Keyphrases
- computational cost
- convergence rate
- neural network
- theoretical justification
- optimization problems
- iterative algorithms
- learning algorithm
- weight update
- convergence analysis
- noisy data
- theoretical analysis
- computational complexity
- multilayer perceptron
- training phase
- computationally efficient
- worst case
- stochastic approximation
- significant improvement
- general regression neural networks