嵌入式隐Markov模型的分段训练方法

Segmental training scheme for embedded hidden markov model

针对嵌入式隐Markov模型再学习问题,提出了分段训练方法用于人脸识别:把当前的训练样本看作整体训练样本的一部分,训练结束后存储训练后的模型参数和中间变量;增加新样本后,以当前模型参数作为初始模型参数,用新增样本训练模型,得到新的中间变量,最后将已存储的中间变量和用新样本计算得到的中间变量合成,得到最终的模型.人脸识别实验结果表明了该方法的可行性.

A segmental scheme to retrain E-HMM （embedded hidden Markov models） for face recognition was presented. The current samples were assumed to be a subset of the whole training samples, after the training process, the E-HMM parameters and the necessary temporary parameters in the parameter re-estimating process were. saved for the use of next step. When new training samples were added, the trained E-HMM parameters were chosen as the initial parameters, the E-HMM was retrained based on the new samples and the new temporary parameters were obtained. These temporary parameters were combined with the saved temporary parameters to form the final E-HMM parameters so that one person face was presented. Experiments on face database showed that the segmental training method was effective.