PAT: Parallel Attention Transformer for Visual Question Answering in Vietnamese
This addresses visual question answering for Vietnamese speakers, representing an incremental improvement with domain-specific application.
The authors tackled visual question answering in Vietnamese by proposing a Parallel Attention Transformer with a Hierarchical Linguistic Features Extractor, achieving the best accuracy on the ViVQA benchmark compared to all baselines and SOTA methods.
We present in this paper a novel scheme for multimodal learning named the Parallel Attention mechanism. In addition, to take into account the advantages of grammar and context in Vietnamese, we propose the Hierarchical Linguistic Features Extractor instead of using an LSTM network to extract linguistic features. Based on these two novel modules, we introduce the Parallel Attention Transformer (PAT), achieving the best accuracy compared to all baselines on the benchmark ViVQA dataset and other SOTA methods including SAAA and MCAN.