Session B, Poster 25: Learning For Integer-Constrained Optimization Through Neural Networks With Limited Training