最大-乘积模糊联想记忆的神经网络学习算法

A Novel Neural Network Learning Algorithm for Max product Fuzzy Associative Memories

在模糊系统中 ,从某种意义上说 ,乘积关系编码可以比最小关系编码保留更多的信息 .提出了最大乘积模糊联想记忆的一种新的神经网络学习算法 ,并给出了严格的理论证明 .该算法能够将多个模糊模式对可靠地编码存储到尽可能少的连接权矩阵中 ,从而大大地减少存储空间 ,而且容易实现 ,并举例验证了它的有效性 .

In fuzzy systems,in certain sense the correlation product encoding preserves more information than the correlation min encoding.A novel neural network learning algorithm for max product fuzzy associative memories(FAMs)is presented with its strict theoretical proofs.Multiple fuzzy pattern pairs can be encoded to store in the FAM connection weight matrixes as few as possible by the algorithm,so it can cut down memory size greatly and this algorithm can be carried out easily.Its effectiveness is testified by an example.