Minimum classification error (MCE) rate method is the most straightforward criterion for HMM training. Inprinciple, it is much better than the maximum likelihood method.
最小错识率(MCE) HMM训练方法是最直接的判决训练方法之一,原理上比最大似然接方法优越得多。
This paper presented a optimum linear multiuser detector in terms of minimum mean bit error rate, and given a approximate method of solving the detector - method of training single perceptron.
本文提出了平均误比特率最小意义下的最佳线性多用户信号检测器,并给出了求解这种最佳线性多用户信号检测器的近似方法——训练单层感知器法。
应用推荐