BUS:Efficient and Effective Vision-language Pre-training with Bottom-Up Patch Summarization
This addresses a key bottleneck in vision-language pre-training for researchers and practitioners by improving efficiency and effectiveness, though it is incremental as it builds on existing VLP methods.
The paper tackles the inefficiency and ineffectiveness of Vision Transformer (ViT) based Vision-Language Pre-training (VLP) models due to lengthy visual token sequences by proposing BUS, a Bottom-Up Patch Summarization approach that coordinates extraction and abstraction to learn concise summaries, resulting in a 50% boost in training efficiency and state-of-the-art performance on many downstream tasks without increasing computational costs.
Vision Transformer (ViT) based Vision-Language Pre-training (VLP) models have demonstrated impressive performance in various tasks. However, the lengthy visual token sequences fed into ViT can lead to training inefficiency and ineffectiveness. Existing efforts address the challenge by either bottom-level patch extraction in the ViT backbone or top-level patch abstraction outside, not balancing training efficiency and effectiveness well. Inspired by text summarization in natural language processing, we propose a Bottom-Up Patch Summarization approach named BUS, coordinating bottom-level extraction and top-level abstraction to learn a concise summary of lengthy visual token sequences efficiently. Specifically, We incorporate a Text-Semantics-Aware Patch Selector (TSPS) into the ViT backbone to perform a coarse-grained visual token extraction and then attach a flexible Transformer-based Patch Abstraction Decoder (PAD) upon the backbone for top-level visual abstraction. This bottom-up collaboration enables our BUS to yield high training efficiency while maintaining or even improving effectiveness. We evaluate our approach on various visual-language understanding and generation tasks and show competitive downstream task performance while boosting the training efficiency by 50\%. Additionally, our model achieves state-of-the-art performance on many downstream tasks by increasing input image resolution without increasing computational costs over baselines.