CAJ | 학술논문

为了得到似然函数不解析可得的 HMM 隐状态估计，将HMM 隐状态估计看成一个贝叶斯最优滤波问题，采用基于近似贝叶斯计算的离子滤波算法对此类问题进行求解，从而解决了一些常用算法如卡尔曼滤波、扩展卡尔曼滤波及离子滤波等都不能解决的似然函数不解析可得的滤波问题。
위료득도사연함수불해석가득적 HMM 은상태고계，장HMM 은상태고계간성일개패협사최우려파문제，채용기우근사패협사계산적리자려파산법대차류문제진행구해，종이해결료일사상용산법여잡이만려파、확전잡이만려파급리자려파등도불능해결적사연함수불해석가득적려파문제。
In order to obtain the estimation of HMM's hidden state when the likelihood function of HMM is not analytically available ,we view the estimation of HMM's hidden state as a bayesian optimal filtering problem ,and adopt the particle filter algorithm based on approximate Bayesian computation to resolve such problem .As a result ,we solve the filtering problem which likelihood function is not analytically available and some of other methods like the Kalman filter or extend Kalman filtering and particle filter can't solve .