Modified Model-Agnostic Meta-Learning
Aashay Pawar · 2020
Meta-learning, an idea of "learning to learn," is a machine learning field that applies a learning algorithm to train a model for performing various tasks. This paper extends the current version of meta-learning by applying a modified model-agnostic algorithm so that the model becomes capable of performing the tasks just upon being trained on a few shots of samples. Moreover, the proposed algorithm also helps facilitate problems like over-fitting.