一种协同神经网络不平衡参数的差分优化方法

An Unbalanced Parameters Optimization of Synergetic Neural Network based on Differential Evolution Algorithm

协同模式识别方法是协同学原理应用于模式识别领域的一种新方法,识别过程中通过调整神经网络的注意参数,能改进系统的识别性能,对协同神经网络参数在不平衡条件下进行优化,能充分利用协同神经网络的自学习能力,以提高识别效果。差分进化作为一种有效的全局近似最优解的搜索算法,具有收敛性好、速度快的特点,文中提出了一种基于差分进化的方法对协同神经网络中参数进行优化,在协同神经网络的参数空间搜索最优参数,采用了均方适应度方差的机制自适应调整搜索速度和搜索精度,克服差分进化算法参数调整困难的不足,以提高算法的寻优能力,新方法具有全局兼局部寻优能力,不易陷入局部极值,同时新方法采用约简的序参量进化参数,使优化算法能有效提高协同神经网结的效率,实际图像的分类识别结果表明,注意参数的变化会导致完全不同的识别结果,另外,本文还将新算法与平衡参数的方法、其它优化的非平衡参数的协同学习算法进行了全局优化能力的比较,采用新方法具有更快的收敛速度和更优的分类识别效果。

Synergetic pattern recognition method is a novel approach to pattern recognition based on synergetic principles.We could improve recognition performance of SNN（Synergetic Neural Network） by adjusting attention parameters in the neural network system. The unbalanced parameters of Synergetic neural network are optimized to improve the recognition effect by sufficiently using the self-learning abilities of SNN.Differential evolution,as an effective searching algorithm for global approximate optimal solution,has the characteristics of convergence fast to better solution.In this paper,an algorithm of parameters optimization based on differential evolution was proposed.This new algorithm is used to search the global optimum attention parameters of SNN in the corresponding parameter space.Fitness mean square variance is adopted to modify searching speed and searching precision in the adaptive manner,because the parameters of differential evolution algorithm are hard to adopt dynamically.and the way of fitness mean square variance could helps improving the optimizing abilities of the algorithm.The new algorithm has better parameter searching abilities,both globally and locally, and can hardly been trapped into local extreme.A simplified version of the order parameter set is applied in this algorithm which improves the recognition ratio of the SNN system effectively.Experiment result on real images show that the change of attention parameters could lead to completely different recognition results.Comparisons with the algorithm under balanced attention parameters and with other optimization algorithm under unbalanced attention parameters are made and the proposed algorithm has faster convergent speed and better recognition result.