An Approximation Algorithm to Find the Largest Linearly Separable Subset of Training Examples
Mario Marchand, Mostefa Golea · 1993
We present an approximation algorithm for the NP-hard problem of finding the largest linearly separable subset of examples among a training set. The algorithm uses, incrementally, a Linear Programming procedure. We present numerical evidence of its superiority over the Pocket algorithm used by many neural net constructive algorithms.