FW-based SVMs
This is the C++ code used in [6,8,9]. The structure of the code is based on the LIBSVM (Lin et al., 2002) and LIBCVM (Tsang et al., 2005) packages, with a very similar interface.
It implements five different Frank-Wolfe solvers for SVM classification problems (FW, Modified FW, Pairwise/SWAP FW, PARTAN FW, Fully Corrective FW), and offers the option of performing approximate iterations with a stochastic speed-up technique.
Main developer: Ricardo Nanculef (UTFSM Chile)
Download link (last update: 21/01/2015)
FW-based LASSO
This C++ code implements a stochastic Frank-Wolfe algorithm for solving large-scale LASSO regression problems.
Main developer: Ricardo Nanculef (UTFSM Chile)
Download link (last update: 10/04/2015)