Reducing Safety Interventions in Provably Safe Reinforcement Learning | IEEE Conference Publication | IEEE Xplore