机译:通过分级注意机制和策略梯度优化进行图像字幕
Queens Univ Belfast Sch Elect Elect Engn & Comp Sci Belfast Antrim North Ireland;
Inst Adv Artificial Intelligence Nanjing Nanjing Jiangsu Peoples R China|Horizon Robot Beijing Peoples R China|Chinese Acad Sci Inst Automat Beijing Peoples R China|East China Normal Univ Sch Comp Sci & Software Engn Shanghai Peoples R China;
Univ Liverpool Elect Engn & Elect Liverpool Merseyside England|Xian Jiaotong Liverpool Univ Dept Comp Sci & Software Engn Suzhou Peoples R China;
Univ Liverpool Elect Engn & Elect Liverpool Merseyside England;
Xian Jiaotong Liverpool Univ Dept Comp Sci & Software Engn Suzhou Peoples R China;
Inst Adv Artificial Intelligence Nanjing Nanjing Jiangsu Peoples R China|Horizon Robot Beijing Peoples R China;
Image captioning; Hierarchical attention mechanism; Generative adversarial network; Reinforcement learning; Policy gradient;
机译:告诉和猜测:用于自然图像标题的合作学习,具有分层精致的注意力
机译:基于分层多模式注意力的神经网络的图像字幕
机译:具有双重关注机制的图像字幕生成
机译:通过蜘蛛的政策梯度优化改进了图像标题
机译:一种新的分层多尺度优化方法:用于水驱优化的梯度和非梯度方法。
机译:多型网:基于多思科领域的残差模块和基于VHR图像语义分割的优化U-Net
机译:通过spIDEr的策略梯度优化改进了图像标题