SOM网络即自组织特征映射网络，采用竞争学习规则——Winner-Take-All 。网络的输出神经元之间相互竞争以求被激活，结果在每一时刻只有一个输出神经元被激活。这个被激活的神经元称为竞争获胜神经元，而其它神经元的状态被抑制，故称为Winner Take All。SOM网络是以生物学为基础，其运行分为训练和工作两个阶段。-SOM network self- organizing feature map network, competitive learning rule- Winner- Take-All . Competition between the output neurons of the network in order to be activated, results in every moment only one output neuron is activated . This activation of neurons called the competition winning neuron , while the other neuron　s state is suppressed , it is known as Winner Take All . The SOM network is biologically based , the run is divided into training and work in two stages .