APE: Aligning Pretrained Encoders to Quickly Learn Aligned Multimodal Representations
This work addresses the challenge of reducing training costs and data requirements for multimodal alignment, which is significant for researchers and practitioners in AI seeking efficient and robust models, though it is incremental as it builds on existing pretrained encoders.
The paper tackles the problem of learning aligned multimodal representations efficiently by aligning pretrained unimodal encoders with small auxiliary functions, achieving competitive or superior performance to state-of-the-art methods while using two orders of magnitude less time and data and training 77% fewer parameters, as demonstrated by surpassing prior state of the art for ImageNet zero-shot classification.
Recent advances in learning aligned multimodal representations have been primarily driven by training large neural networks on massive, noisy paired-modality datasets. In this work, we ask whether it is possible to achieve similar results with substantially less training time and data. We achieve this by taking advantage of existing pretrained unimodal encoders and careful curation of alignment data relevant to the downstream task of interest. We study a natural approach to aligning existing encoders via small auxiliary functions, and we find that this method is competitive with (or outperforms) state of the art in many settings while being less prone to overfitting, less costly to train, and more robust to distribution shift. With a properly chosen alignment distribution, our method surpasses prior state of the art for ImageNet zero-shot classification on public data while using two orders of magnitude less time and data and training 77% fewer parameters.