Abstract:
Multilayer feedforward networks are one
of the most used neural networks in various
domains because of their universal approximation
ability. One of the popular algorithms for training
multilayer feedforward network is
backpropagation which uses two phase namely
feedforward and backpropagate to learn the
weight in the network. The main disadvantage of
the backpropagation algorithm is its convergence
rate is slow at it always being trapped in local
minima.Artificial Bee Colony (ABC) algorithm is
one of the most recently introduced swarmbasedalgorithms.
ABC simulates the intelligent
foraging behavior of a honeybee swarm.The
proposed method in this paper includes an
artificial bee colony algorithm based neural
network training method and back-propagation
based neural network. And then compare accuracy
and mean square rate for both neural network
training. Four type of UCI datasets are used for
both neural network training.
