Bounded PSO Vmax Function in Neural Network Learning
Y. S. Lee, Siti Mariyam Shamsuddin, Haza Nuzly Abdull Hamed · 2008
Typically, back propagation (BP) algorithm is the most widespread technique in Artificial Neural Network (ANN learning). However, major disadvantages of BP are due to its convergence rate sluggishness and always being trapped at the local minima. Consequently, Particle Swarm Optimization(PSO) is chosen and applied in feed forward neural network to enhance the network learning. In conventional PSO, maximum velocity Vmax is exploited to serve as a constraint that controls the maximum global exploration ability of PSO. By setting these values too small cause the limitation of maximum global exploration. Hence, PSO will always favor for a local search regardless the values of weight inertia. However, by setting to a large maximum velocity, PSO can have a large range of exploration ability. Therefore, in this study, we proposed different bounded functions of PSO Vmax to control the global exploration of particles. The results show that bounded Vmax of hyperbolic tangent function furnish promising outcomes compared to bounded Vmax sigmoid function and standard Vmax function.