Kernel-Free Quadratic Surface Support Vector Regression with Non-Negative Constraints

In this paper, a kernel-free quadratic surface support vector regression with non-negative constraints (NQSSVR) is proposed for the problem. The task of NQSSVR to find function as function. By utilizing technique, model avoids difficulty choosing kernel and corresponding parameters, has interpretability certain extent. fact, data may have priori information that value response variable will increase explanatory grows in interval. Moreover, order ensure monotonically increasing on interval, respect coefficients are introduced construct optimization problem NQSSVR. And obtained by matches information, which been proven theoretical analysis. addition, existence uniqueness solution primal dual NQSSVR, relationship between them addressed. Experimental results two artificial datasets seven benchmark validate feasibility effectiveness our approach. Finally, method verified real examples air quality.