Efficient Vision-Language Pre-training by Cluster Masking
This work addresses efficiency and representation quality in vision-language models, though it appears incremental as it builds on existing masking techniques.
The paper tackles the problem of improving vision-language pre-training by introducing a cluster masking strategy that masks visually similar image patches, which enhances representation quality and speeds up training. It outperforms other masking methods like FLIP on benchmark evaluations.
We propose a simple strategy for masking image patches during visual-language contrastive learning that improves the quality of the learned representations and the training speed. During each iteration of training, we randomly mask clusters of visually similar image patches, as measured by their raw pixel intensities. This provides an extra learning signal, beyond the contrastive training itself, since it forces a model to predict words for masked visual structures solely from context. It also speeds up training by reducing the amount of data used in each image. We evaluate the effectiveness of our model by pre-training on a number of benchmarks, finding that it outperforms other masking strategies, such as FLIP, on the quality of the learned representation.