MAML

This repository contains files relating to my MSc Data Science research project Model-Agnostic Meta-Learning for Few-Shot Deep Learning. This thesis focused on researching MAML to solve few-shot deep learning problems. To these ends, Iconducted a thorough literature review on deep learning and meta-learning, focusing on meta-learningalgorithms MAML and MAML++. The project involved implementing two experimental studies using Python; a toy problem demonstrating the effectiveness of MAML in addition to implementing MAML on a computer vision problem

Files

1. Model_Agnostic_Meta_Learning_for_Few_Shot_Deep_Learning.pdf

• PDF file containing submitted thesis for project.

3. MAML_PyTorch.ipynb

• Python code demonstrating the effectiveness of MAML through a toy problem

5. CIFAR_FS_maml.ipynb

• Python code implementing MAMLon a computer vision problem.