.png){kind=icon}
.png){kind=icon}
Return
Linear convergence rate for the MDM algorithm for the Nearest Point Problem
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2014.10.015.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In this paper we will prove a linear convergence rate for the extension of the Mitchell, Dem'yanov and Malozemov (MDM) algorithm for solving the Nearest Point Problem (NPP). While linear convergence proofs for the related (but different) SMO method intended for SVM training require that the kernel matrix be positive definite, no such assumption is needed in NPP for MDM. Moreover, we will also show linear convergence for the sequence of MDM vectors to the unique solution vector W* of NPP and for a quantity that measures the gap in the Karush-Kuhn-Tucker conditions. Furthermore, even if there might be several multiplier representations for le, we will show that any MDM-generated multiplier sequence converges linearly to an optimal multiplier. This linear convergence is shown to be optimal and it is also numerically illustrated over six datasets. We will follow an approach that relies on a geometric point of view that yields a simple path to the proofs. (C) 2014 Elsevier Ltd. All rights reserved.
Keywords:
Convergence
MDM algorithm
Nearest Point Problem
Convex Hulls
Support vector machines
nu-SVMs
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.6
Papers:
1.3W
Citations:
4.5W
Organization
No organization information available