Efficient learning in neural networks
Date of Issue1998
School of Electrical and Electronic Engineering
In studies of neural networks, the Multilavered Feedforward Network is the most widely used network architecture while the Backpropagation (BP)algo-rithm is the prime learning algorithm for this kind of networks. Although the BP algorithm is thought to be based on a solid theoretical background, some of its drawbacks hamper its use in solving problems efficiently. These draw-backs include slow convergence, the problem of local minima, and the degra-dation in performance due to practical constraints such as limited weight precision in real implementation of the algorithm.
DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems
NANYANG TECHNOLOGICAL UNIVERSITY