This paper evaluates alternative measures of core inflation in a multisectoral New Keynesian model. Stabilizing sticky-price inflation is considered optimal; however, there are different measures of rigid price inflation in a multi-sector economy. Central banks commonly monitor limited influence estimators that exclude volatile sectors price changes. This would provide a measure of core that mostly correlates to the rigid sectors of the economy, and consequently, could provide welfare gains. Based on the multi-sector model of Carvalho, Lee e Park (2021), a proposed modification is considered: an interest rate rule that responds to a measure of core inflation instead of the usual headline. This modification introduces a non-linearity we tackle using frontier machine learning methods. We calibrate sector-specific supply shocks through indirect inference to match the moments observed in CPI sectoral inflation. Our results show that responding to alternative core measures can reduce consumer welfare losses.