【文献阅读】seada-VQA对数据进行对抗增强并保留语义正确性(R. Tang等人,ArXiv,2020)
一、背景文章题目:《SemanticEquivalentAdversarialDataAugmentationforVisualQuestionAnswering》文章下载地址:https://arxiv.org/pdf/2007.09592.pdf文章引用格式:RuixueTang,ChaoMa,WeiEmmaZhang,QiWu,andXiaokangYang."SemanticEquival