给定观测序列 O=O1O2O3…Ot和模型参数λ=(A,B,π)，怎样有效计算某一观测序列的概率，进而可对该HMM做出相关评估。例如，已有一些模型参数各异的HMM，给定观测序列O=O1O2O3…Ot，我们想知道哪个HMM模型最可能生成该观测序列。通常我们利用forward算法分别计算每个HMM产生给定观测序列O的概率，然后从中选出最优的HMM模型。-Given the observation sequence O = O1O2O3 ... Ot and model parameters λ = (A, B, π), how to effectively calculate the probability that a single observation sequence, and thus can make the relevant assessment of the HMM. For example, there are a number of different model parameters HMM, given the observation sequence O = O1O2O3 ... Ot, we want to know which model is most likely to generate the HMM observation sequence. Usually we use forward algorithm calculates the probability of a given observation sequence O generated for each HMM, HMM and then to choose the best model.