论文部分内容阅读
本文对照经典的矢量量化算法的不足,讨论了基于竞争机制的连续Hopfield神经网络矢量量化算法的设计与实现。详细描述了网络映射过程、网络能量函数的刻画和神经元状态转换方程。实验结果表明,与经典的LBG算法相比,本文所提算法具有更好的性能和强大的并行处理能力以及更优良的全局优化能力。
In this paper, the shortage of classical vector quantization algorithm is discussed. The design and implementation of continuous Hopfield neural network vector quantization algorithm based on competitive mechanism are discussed. Described in detail the network mapping process, characterization of network energy function and neuron state transition equations. Experimental results show that compared with the classical LBG algorithm, the proposed algorithm has better performance and powerful parallel processing ability and better global optimization ability.