Unsupervised Classification via Convex Absolute Value Inequalities
نویسنده
چکیده
We consider the problem of classifying completely unlabeled data by using convex inequalities that contain absolute values of the data. This allows each data point to belong to either one of two classes by entering the inequality with a plus or minus value. By using such absolute value inequalities (AVIs) in support vector machine classifiers, unlabeled data can be successfully partitioned into two classes that capture most of the correct labels dropped from the data. Inclusion of partially labeled data leads to a semisupervised classifier. Computational results include unsupervised and semisupervised classification of the Wisconsin Breast Cancer Wisconsin (Diagnostic) Data Set.
منابع مشابه
Unsupervised and Semisupervised Classification Via Absolute Value Inequalities
We consider the problem of classifying completely or partially unlabeled data by using inequalities that contain absolute values of the data. This allows each data point to belong to either one of two classes by entering the inequality with a plus or minus value. By using such absolute value inequalities (AVIs) in linear and nonlinear support vector machines, unlabeled or partially labeled data...
متن کاملOn Fejér Type Inequalities for (η1,η2)-Convex Functions
In this paper we find a characterization type result for (η1,η2)-convex functions. The Fejér integral inequality related to (η1,η2)-convex functions is obtained as a generalization of Fejér inequality related to the preinvex and η-convex functions. Also some Fejér trapezoid and midpoint type inequalities are given in the case that the absolute value of the derivative of considered function is (...
متن کاملOn new inequalities of Hermite-Hadamard-Fejér type for convex functions via fractional integrals
In this paper, we establish some weighted fractional inequalities for differentiable mappings whose derivatives in absolute value are convex. These results are connected with the celebrated Hermite-Hadamard-Fejér type integral inequality. The results presented here would provide extensions of those given in earlier works.
متن کاملSome Hermite-hadamard Type Inequalities for Functions Whose Partial Derivatives in Absolute Value Are Preinvex on the Co-ordinates
In this paper we point out some inequalities of Hermite-Hadamard type for double integrals of functions whose partial derivatives in absolute value are preinvex on the co-ordinates on rectangle from the plane. Our established results generalize some recent results for functions whose partial derivatives in absolute value are convex on the co-ordinates on the rectangle from the plane.
متن کاملInequalities of Ando's Type for $n$-convex Functions
By utilizing different scalar equalities obtained via Hermite's interpolating polynomial, we will obtain lower and upper bounds for the difference in Ando's inequality and in the Edmundson-Lah-Ribariv c inequality for solidarities that hold for a class of $n$-convex functions. As an application, main results are applied to some operator means and relative operator entropy.
متن کامل