Title: Correcting for Leakage Energy in the SiD Silicon-Tungsten ECal

Abstract: A dominant contribution to ECal resolution at high energy (eg. 100 GeV) comes from leakage beyond the containment of the calorimeter. We have studied the leakage energy for the SiD silicon-tungsten ECal and developed a neural network algorithm for estimating the leakage energy and correcting the energy measurement. The SiD TDR design calls for 20 thin 2.5 mm tungsten layers followed by 10 thick 5.0 mm tungsten layers. We have investigated the impact on the leakage energy of a reduced number of layers, and the ability of an optimized neutral network analysis to correct for the leakage with a reduced number of layers, and reduced material thickness. Reducing layer numbers is motivated by cost containment.