| 联想记忆网络中的最速递减动态规则 |
| |
| 引用本文： | 汪涛 俞瑞钊. 联想记忆网络中的最速递减动态规则[J]. 计算机学报, 1995, 18(6): 475-479 |
| |
| 作者姓名： | 汪涛 俞瑞钊 |
| |
| 作者单位： | 浙江大学计算机科学与工程系 |
| |
| 摘 要： | 本文提出了一种联想记忆网络中的最速递减动态演化规则，首先改变能产生最大系统能量递减的神经元状态，从而减小落入多余吸引子的可能性。在理论上，我们分析了样本的吸引域和联想规则的收敛性，计算机实验结果说明了它的优越性。
|
| 关 键 词： | 联想记忆网络 最速递减 多余吸引子 吸引域 |
STEEPEST DESCENT DYNAMICS FOR ASSOCIATIVE MEMORIES |
| |
| Wang Tao, Yu Ruizhao and He Zhijun. STEEPEST DESCENT DYNAMICS FOR ASSOCIATIVE MEMORIES[J]. Chinese Journal of Computers, 1995, 18(6): 475-479 |
| |
| Authors: | Wang Tao Yu Ruizhao He Zhijun |
| |
| Abstract: | This paper,gives a steepest descent dynamics for associative memones. To reduce the probability of falling into spurious states, the dynamics first updates those unit states that can produce larger energy decrements. Basins of attraction of stored patterns and convergence of the dynamics are studied theoretically. Computer simulations have shown its efficiency. |
| |
| Keywords: | Associative memory steepest descent spurious state basin of attraction. |
| 本文献已被 CNKI 维普 等数据库收录！ |
