论文部分内容阅读
混合高斯模型能够有效地拟合概率密度函数,常用的混合高斯概率密度模型参数估计方法是EM迭代算法,这种算法的缺点是估计精度过分依赖于初始值,而且不能估计模型阶数。基于遗传算法的K-means初始化EM算法可以同时估计模型阶数和参数。试验结果表明,该算法具有更好的聚类效果。
The Gaussian mixture model can effectively fit the probability density function. The common parameter estimation method for the Gaussian mixture density model is the EM iterative algorithm. The disadvantage of this algorithm is that the estimation accuracy is too dependent on the initial value and the model order can not be estimated. K-means initialization EM algorithm based on genetic algorithm can estimate model order and parameters simultaneously. Experimental results show that this algorithm has a better clustering effect.