MMCL

Experiments for the paper "Understanding Multimodal Contrastive Learning and Incorporating Unpaired Data"

1. For the experiment in Section 5.1, see MMCL folder
2. For the experiment in Section 5.2, see self-supervised MMCL/synthetic data
3. For the experiment in Section 5.3, see self-supervised MMCL/real data

[Reference]

1. Understanding Multimodal Contrastive Learning and Incorporating Unlabeled Data, R. Nakada, I. Gulluk, Z. Deng, W. Ji, J. Zou, L. Zhang, The 26th International Conference on Artificial Intelligence and Statistics (to appear), 2023
2. Convolutional Autoencoder in Pytorch on MNIST dataset