Theoretical Corrections and the Leveraging of Reinforcement Learning to Enhance Triangle Attack