| 초록 |
One of the most challenging and aspiring aims of computational systems biology is the framework of quantitative prediction with the assistance of mathematical models. However, with the model nonlinearity getting higher as well as the larger number of parameters to be estimated simultaneously, the traditional optimization methods are not as efficient as before. In this work, the Particle Swarm Optimization (PSO) method is used for the estimation of model parameters in highly nonlinear metabolic networks in systems biology. PSO isa novel metaheuristic optimization methods, and with the modification of the essential parameters changing strategy, the convergence speed of the hybrid PSO has been accelerated. In this work, the comparison of performances between conventional deterministic methods, stochastic methods and PSO is studied. Also, the parallelization of PSO is also applied to achieve the enhancement of the performance. It is shown that the suggested PSO-based method is capable of minimizing the objective function better and faster and has robustness of estimating the model parameters more successfully. |
