Image Captioning via Compact Bidirectional Architecture
通过紧凑双向架构实现图像描述
机构 * School of Big Data and Statistics, Anhui University(安徽大学大数据与统计学院) ; Institute of Artificial Intelligence, Hefei Comprehensive National Science Center(合肥综合性国家科学中心人工智能研究院) ; Intelligent Interconnected Systems Laboratory of Anhui Province (Hefei University of Technology)(安徽省智能互联系统实验室(合肥工业大学)) ; School of Computer Science and Information Engineering, Hefei University of Technology(合肥工业大学计算机与信息学院) ; JD Explore Academy, JD.com Inc(京东探索研究院) ; State Key Laboratory of Digital Intelligent Technology for Unmanned Coal Mining, Anhui University of Science and Technology(安徽理工大学深部煤矿采动响应与灾害防控国家重点实验室)
专题命中 多模态RAG :retriever(abstract);分类 cs.CL
AI总结 本文提出紧凑双向Transformer模型,通过并行解码器隐式和显式利用双向上下文,经消融研究发现其比显式交互机制更重要,结合词级融合进一步提升效果,并扩展至LSTM基座实现新SOTA结果。