LARS: efficient algorithm to solve lasso

Least angle regression

elastic net: in case of collinear dictionary atoms, it will pick collinear atoms together, or drop them together.

group lasso: in the extreme case, it behavors like lasso, however in a group manner, i.e. a group is either picked or dropped, and sparsity is obtained at the group level, but within each group, sparsity can't be guarantteed.

sparse group lasso: sparsity is gotten at both group level and within each group.

adaptive lasso:

Packages:

(1) LARS: l1-magic http://statweb.stanford.edu/~candes/l1magic/

(2) CRAN - Package glmnet, Package 'glmnet'

(3) SPAMS: http://spams-devel.gforge.inria.fr/

(4) SLEP: http://www.public.asu.edu/~jye02/Software/SLEP/

(5) SIRS: http://www-bcf.usc.edu/~jinchilv/publications/software/