A geometrical analysis of associative memory
Neural Networks
ICNC '09 Proceedings of the 2009 Fifth International Conference on Natural Computation - Volume 01
Hi-index | 0.00 |
In this paper an efficient graph coloring algorithm based on the transiently chaotic neural network (TCNN) is presented. This algorithm apply the TCNN with hysteretic output function instead of logistic output function, this make the model has higher ability of overcoming drawbacks that suffer from the local minimum. Meanwhile, a rapid strategy is merged in this model in order to avoid oscillation and offer a considerable acceleration of converging to the optimal solution. The numerical simulation results demonstrated that the proposed model has higher ability and more rapid speed to search for globally optimal solution of the graph coloring problem than the previous TCNN model with logistic output function and without the rapid strategy.