Curriculum Learning for Data-Efficient Vision-Language Alignment
This work addresses the data inefficiency challenge in vision-language alignment for researchers and practitioners, though it is incremental as it builds on existing pre-trained models and contrastive learning methods.
The paper tackles the problem of aligning image and text encoders with limited paired data by aligning pre-trained models using a curriculum learning algorithm, achieving better zero-shot image retrieval performance than CLIP while using less than 1% of the training data.
Aligning image and text encoders from scratch using contrastive learning requires large amounts of paired image-text data. We alleviate this need by aligning individually pre-trained language and vision representation models using a much smaller amount of paired data, augmented with a curriculum learning algorithm to learn fine-grained vision-language alignments. TOnICS (Training with Ontology-Informed Contrastive Sampling) initially samples minibatches whose image-text pairs contain a wide variety of objects to learn object-level alignment, and progressively samples minibatches where all image-text pairs contain the same object to learn finer-grained contextual alignment. Aligning pre-trained BERT and VinVL models to each other using TOnICS outperforms CLIP on downstream zero-shot image retrieval while using less than 1% as much training data.