Detached breakwaters are critical coastal protection measures, yet conventional emerged designs often have significant environmental and aesthetic impacts. Submerged and low-crested breakwaters offer a promising alternative, but the morphodynamic responses they induce remain poorly understood, particularly in the formation of tombolos and salients—key sedimentary features. Existing definitions of these formations often rely on dimensionless ratio analysis, which can lead to contradictions and inconsistencies depending on the selected methodology. While prior studies have predominantly focused on fully emerged structures morphodynamics, this study addresses the uncertainty surrounding tombolo and salient predictions for submerged and low-crested structures. A novel framework is proposed, utilizing Machine Learning (ML) classification techniques as an alternative to traditional methodologies, aiming to improve the reliability of predictive models and reduce uncertainty in the formation limits of tombolo and salient events.