Global optimization of neural network weights using subenergy tunneling function and ripple search
Ye Hong, Zhiping Lin · 2003
This paper presents a new approach to supervised training of weights in multilayer feedforward neural networks. The algorithm is based on a subenergy tunneling function to reject searching in unpromising regions and a ripple-like global search to get away from local minima. The global convergence properties of the proposed algorithm are demonstrated through three frequently used neural network learning applications. The performance of the new technique is better than or at least similar to that of other training methods in the literature. The proposed method is flexible and conceptually simple to implement.