演講公告 89年5月25日 (星期四)

The applications of acyclic phase-type distributions on semi-Markov decision processes

摘要

It is well-known that the main barrier in analyzing semi-Markovian model is that the mathematical formulation is so complicated that it is essentially intractable. To overcome this barrier, an acyclic phase-type approach is proposed to approximate the general sojourn time distributions of a semi-Markov process. Using the approximation, a semi-Markovian model can be transformed into a Markovian model such that the analytical tractability of Markov processes can be preserved. Based on the resulting Markovian model, algorithms are provided to derive the optimal state-dependent and state-age-dependent policies. Furthermore, procedures are developed to implement these optimal policies back on the original semi-Markovian model.