Parameter Estimation of Silicon Metal Grid using Supervised Learning

Silicon industry needs reduced design time to cater to broad annual product portfolio. Therefore, avoiding complex simulations during product design has immense value. To that end this paper presents machine learning based parameter estimation method for silicon metal grid based on past data. Regression results from employed machine learning algorithms and dependency on data standardization is discussed. Over 40% reduction in root mean square error of grid resistance is reported which is crucial for obtaining accurate transient and AC simulation result.