Many engineering and economic applications can be formulated by a minimization problem subject to a single linear constraint and bounds on the variables and the majority of them are carachterized by very large- scale dimensions; we present a new algorithm that, combining a Decom- position strategy with a Truncated Newton approach, allows to solve this type of problems efficiently. In particular we applied the algorithm to the solution of problems arising in the training of Support Vector Machines, whose dual formulation is carachterized by this particular structure.
A New Algorithm for the Solution of Large-Scale Singly Linearly Constrained Problems Subject to Simple Bounds / Liuzzi, Giampaolo; Lucidi, Stefano; Manno, Andrea; Francesco, Rinaldi. - STAMPA. - (2013). (Intervento presentato al convegno 59th Workshop Nonlinear Optimization: a Bridge from Theory to Applications tenutosi a Erice nel 10-17/6/2013).
A New Algorithm for the Solution of Large-Scale Singly Linearly Constrained Problems Subject to Simple Bounds.
Giampaolo Liuzzi;LUCIDI, Stefano;MANNO, ANDREA;
2013
Abstract
Many engineering and economic applications can be formulated by a minimization problem subject to a single linear constraint and bounds on the variables and the majority of them are carachterized by very large- scale dimensions; we present a new algorithm that, combining a Decom- position strategy with a Truncated Newton approach, allows to solve this type of problems efficiently. In particular we applied the algorithm to the solution of problems arising in the training of Support Vector Machines, whose dual formulation is carachterized by this particular structure.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.