| L. Wasserman. personal communications. 1999 |
....EM algorithm, the mean, covariance and weight of each Gaussian component can be adjusted freely. However we can fix some parameters to certain values, or limit their ranges during EM training. We call it the restricted EM algorithm. It will still converge to a local maximum in terms of likelihood [15]. More specifically, in this paper we will fix 1 = 1 and limit w 1 to be within range [w low ; w high ] The meaning and value of 1 , w low , and w high will be discussed in next section. By restricting these two parameters, we can enforce N 1 to approximate the hand color distribution. The ....
L. Wasserman. personal communications. 1999
....EM algorithm, the mean, covariance and weight of each Gaussian component can be adjusted freely. However we can fix some parameters to certain values, or limit their ranges during EM training. We call it the restricted EM algorithm. It will still converge to a local maximum in terms of likelihood [15]. More specifically, in this paper we will fix 1 # 1 and limit w 1 to be within range # w low ;w h # . The meaning and value of # 1 , low ,andw hig h will be discussed in next section. By restricting these two parameters, we can enforce N 1 to approximate the hand color distribution. ....
L. Wasserman. personal communications. 1999
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC