What training strategy, named by analogy to education, presents examples in a meaningful order from easy to hard rather than randomly, often improving convergence and final performance?

Curriculum learning

Curriculum learning (Bengio et al., 2009) organizes training so a model masters simpler concepts before harder ones, which can smooth optimization and improve generalization. The idea recurs in reinforcement learning and in staged data mixes for training large models.