A Fast Method for Training Linear SVM in the Primal
نویسندگان
چکیده
We propose a new algorithm for training a linear Support Vector Machine in the primal. The algorithm mixes ideas from non smooth optimization, subgradient methods, and cutting planes methods. This yields a fast algorithm that compares well to state of the art algorithms. It is proved to require O(1/λ ) iterations to converge to a solution with accuracy . Additionally we provide an exact shrinking method in the primal that allows reducing the complexity of an iteration to much less than O(N) where N is the number of training samples.
منابع مشابه
ABS Solution of equations of second kind and application to the primal-dual interior point method for linear programming
Abstract We consider an application of the ABS procedure to the linear systems arising from the primal-dual interior point methods where Newton method is used to compute path to the solution. When approaching the solution the linear system, which has the form of normal equations of the second kind, becomes more and more ill conditioned. We show how the use of the Huang algorithm in the ABS cl...
متن کامل: Primal Estimated sub - GrAdient SOlver for SVM
We describe and analyze a simple and effective stochastic sub-gradient descent algorithm for solving the optimization problem cast by Support Vector Machines (SVM). We prove that the number of iterations required to obtain a solution of accuracy is Õ(1/ ), where each iteration operates on a single training example. In contrast, previous analyses of stochastic gradient descent methods for SVMs r...
متن کاملSome new results on semi fully fuzzy linear programming problems
There are two interesting methods, in the literature, for solving fuzzy linear programming problems in which the elements of coefficient matrix of the constraints are represented by real numbers and rest of the parameters are represented by symmetric trapezoidal fuzzy numbers. The first method, named as fuzzy primal simplex method, assumes an initial primal basic feasible solution is at hand. T...
متن کاملLow-degree Polynomial Mapping of Data for SVM
Non-linear mapping functions have long been used in SVM to transform data into a higher dimensional space, allowing the classifier to separate linearly inseparable data. Kernel tricks are used to handle the huge number of features of the mapped data point. However, the training/testing for large data is often time consuming. Following the recent advances in training large linear SVM (i.e., SVM ...
متن کاملTraining a Support Vector Machine in the Primal
Most literature on support vector machines (SVMs) concentrates on the dual optimization problem. In this letter, we point out that the primal problem can also be solved efficiently for both linear and nonlinear SVMs and that there is no reason for ignoring this possibility. On the contrary, from the primal point of view, new families of algorithms for large-scale SVM training can be investigated.
متن کامل