A Multiple-Criteria and Multiple-Constraint Levels Linear Programming Based Error Correction Classification Model
Beimin Wang, Yunpeng Wang · Procedia Computer Science · 2013
Optimization based classification methods find classifier of a classification problem by solving one or a series of optimization problems, such as linear programming, convex programming and so on. In addition to this, in many applications, for example, credit card account classification, how to handle two types of error is a key issue. Combining linear programming method and error correction, using the structure of a multiple-criteria and multiple-constraint levels linear programming (MC2LP), which allows two alterable cutoffs, two types of error can be systematically corrected. Therefore, in this paper, we pose a new MC2LP model that can correct two types of error. In the new model, two parallel hyperplanes are used to indicate the relative positions between the points and hyperplanes. In other words, it can treat two types of error differently. Especially, it can be developed into a pair of models, which can deal with two types of error respectively. Besides, we will give the matrix representation of the new models and discuss some properties of them. Moreover, experiment result will show the effectiveness of this new dual-hyperplane model.