A novel fuzzy supervised learning method with dynamical parameter estimation for discriminant analysis

In this paper, a reformative supervised fuzzy LDA algorithm (RF-LDA) using a relaxed normalized condition is presented firstly to achieve the distribution information of each sample of images that represented with fuzzy membership degree, which is incorporated into the redefinition of the scatter matrices. Moreover, the need for such a novel fuzzy linear LDA model construction, reduced to parameter estimation when the structure is given beforehand, typically arises when a model is required in order to take some decision about the system, and therefore the dynamical parameter estimation method must recursively process the measured data as they become available. In the line of previous arguments, we approach the problem of controls parameter estimation of RF-LDA by considering the formulation of a HNN, which is named HRF-LDA. Experimental results conducted on the ORL and XM2VTS face databases demonstrate the effectiveness of the proposed method.