【摘要】 基于最大运算Max和t--范数T的神经网络模型Max-T FAM是B.Kosko提出的经典模糊联想记忆(FAM)网络的一种重要的广义形式,其性能有多处不足.本文利用一种参数化聚合算子∨λ,提出了一种计算简单、易于硬件实现的广义模糊联想记忆(GFAM)网络,其连接算子从{∨λ|λ∈[0,1]}中选取;从理论上严格证明了GFAM具有一致连续性,比所有Max-T FAM的映射能力和存储能力强很多;接着运用模糊关系方程理论提出和分析了GFAM的一种所谓的Max-Min-λ学习算法;最后用实验对GFAM和Max-T FAM的完整可靠存储能力进行了比较,并示例了GFAM在图像联想方面的应用.更多还原

【Abstract】 The neural network model Max-T FAM with the maximum operation and a t-norm T is an important gen-eralized form of the classical fuzzy associative memory(FAM) network proposed by B.Kosko.This model has several disadvantages in its properties.Using a parameterized aggregating operator ∨λ,we present a new generalized fuzzy asso-ciative memory(GFAM) network which is simple in computation and easy in implementation by hardware.All conjunctive operators of the interconnections of GFAM are chosen from a cluster {∨λ|λ ∈ [0,1]}.The strict theoretical study reveals that the GFAM is uniformly continuous and has much higher mapping ability and stronger storage capability than all Max-T FAMs.From the theory of fuzzy relational equations,we derive and analyze a so-called Max-Min-λlearning algorithm for GFAM.Experimental comparisons of the storage capability have been made between GFAM and all Max-T FAMs.An application of GFAM to associative images is illustrated.更多还原