Meta-Learning (Learning-to-Learn)

Meta-Learning (Learning-to-Learn) is the capability of an improver algorithm to change its own learning strategies, weight initializations, and hyperparameters.