Features Spaces with Reduced Variables Based on Nearest Neighbor Relations and Their Inheritances.

Generation of useful variables in the features spaces is an important issue throughout the neural networks, the machine learning and artificial intelligence for their efficient and discriminative computations. In this paper, the nearest neighbor relations are proposed for the minimal generation and the reduced variables for the feature spaces. First, the nearest neighbor relations are shown to be minimal independent and inherited for the construction of the feature space. For the analysis, convex cones are made of the nearest neighbor relations, which are independent vectors for the generation of the reduced variables. Then, edges of convex cones are compared for the discrimination of variables. Finally, feature spaces with the reduced variables based on the nearest neighbor relations are shown to be useful for the real documents classification.