What always has been important for researches of the data mining field is inventing of methods to extract the knowledge of high dimension datasets. In this paper, a new method has been recruited to dataset by combination of fuzzy logic, ACO algorithm and genetic algorithm. One of advantages of this method is decreasing of the investigated parameters. For extract of features (reduction of the dimension of datasets), two methods of principal component analyze and fisher's linear discriminant have sequentially recruited at the pre-processing step. In order to evaluate of the proposed method, some datasets from the resource of data mining of UCI which have many characteristics, were selected and were investigated.