论文部分内容阅读
基于ＣＭＡＣ学习过程等价于求解线性方程组的Ｇａｕｓ－Ｓｅｉｄｅｌ迭代这一事实，研究了学习过程的收敛性．利用矩阵分析方法，估计出了收敛的速度．考虑了作为节省存储空间措施的ｈａｓｈ编码的不利影响——破坏了收敛性态．从理论上分析了其存在的原因．
Based on the fact that CMAC learning process is equivalent to Gaus-Seidel iteration of solving linear equations, the convergence of learning process is studied. Using matrix analysis, the rate of convergence is estimated. Considered the detrimental effect of hash encoding as a measure of storage space savings - compromising convergence. From the theoretical analysis of the reasons for its existence.