SparseAA

Sparse Archetypal Analysis

Kayhan Behdin and Rahul Mazumder

Massachusetts Institute of Technology

Introduction

This package obtains a good solution to the Sparse Archetypal Analysis problem,

through Mixed Integer Programming (MIP) based initialization, a block proximal algorithm and a heuristic local search method. The code is written in Julia and a Gurobi license is rquired to use the MIP based initialization.

Usage

For a tutorial, please refer to demo.jl.