 |
Are Random Projections really useful in Mathematical Programming?
1 : LIX CNRS Ecole Polytechnique, Institut Polytechnique de Paris
Centre National de la Recherche Scientifique : UMR7161, Ecole Polytechnique
Random projections are random matrices that can be used to perform dimensionality reduction on vectors. We show how they can be applied to decrease the number of variables and/or constraints of some Mathematical Programming classes, notably Linear Programming, Conic Programming, Quadratic Programming, Quadratically Constrained Quadratic Programming -- and even a convex MINLP encoding a famous clustering problem, and comment on their practical usefulness.
