How to train your MAML
This work addresses practical issues in a widely used meta-learning method for few-shot learning, making it more efficient and robust for researchers and practitioners.
The paper tackles the instability, poor generalization, and high computational cost of Model Agnostic Meta Learning (MAML) in few-shot learning by proposing modifications that stabilize training, improve generalization, and reduce computational overhead.
The field of few-shot learning has recently seen substantial advancements. Most of these advancements came from casting few-shot learning as a meta-learning problem. Model Agnostic Meta Learning or MAML is currently one of the best approaches for few-shot learning via meta-learning. MAML is simple, elegant and very powerful, however, it has a variety of issues, such as being very sensitive to neural network architectures, often leading to instability during training, requiring arduous hyperparameter searches to stabilize training and achieve high generalization and being very computationally expensive at both training and inference times. In this paper, we propose various modifications to MAML that not only stabilize the system, but also substantially improve the generalization performance, convergence speed and computational overhead of MAML, which we call MAML++.