新模糊形态学联想记忆网络的初步研究

人工神经网络理论已经被成功地应用于各种不同的模式识别问题 .重点研究了联想记忆网络 ,提出了一种新的基于形态学和模糊运算的联想记忆网络 ,即模糊形态学联想记忆网络FMAM .它与经典联想记忆和模糊联想记忆FAM有显著不同 .文中分析了FMAM的记忆能力和抗腐蚀 /膨胀噪声的能力 .自联想FMAM具有无限存储能力 ,能保证完全回忆 ,并且回忆在一步内完成 ,可模糊性解释等 .仿真实验验证了自联想FMAM的良好性能

The theory of artificial neural networks has been successfully applied to a wide variety of pattern recognition problems.Associative memories are focused and fuzzy morphological associative memories FMAM based on morphological and fuzzy operations are presented here.The properties of FMAM are greatly different from those of classical associative memories and fuzzy associative memories.FMAM's storage capacity and tolerance capability of erosive/dilative noise are analyzed.Autoassociative FMAM has the unlimited storage capacity,perfect recall guarantee,recall in one step and fuzzy explanation.The experimental results demonstrate the effective performance of autoassociative FMAM.

江苏省青年科技基金 (No BQ2 0 0 0 0 0 3) ;;江苏省自然科学基金 (No BK2 0 0 2 0 0 1 ) ;;中国科学院沈阳自动化研究所机器人学研究室基金(No RL2 0 0 1 0 8) ;;江苏省高校自然科学基金 (No 0 1JKB52 0 0 2 )