Research on BP neural network optimization based on ant colony algorithm

This paper presents an ant colony algorithm and BP algorithm together to complete the learning algorithm for neural networks ACO-BP algorithm. The algorithm adopts the ant colony algorithm for global optimization of the network weights, overcome the disadvantage of BP algorithm that is easy to fall into local optimum; then, the optimal weights found by BP algorithm as the initial value, further optimization. Finally, the simulation experiments show that, if the network structure is determined by the condition, this algorithm not only speeds up the convergence speed of the improved ant colony algorithm of optimal solution, but also can avoid falling into local optimal path. It will increase the reliability.