Attention Correctness in Neural Image Captioning

Chenxi Liu, Junhua Mao, Fei Sha +1

Attention mechanisms have recently been introduced in deep learning for various tasks in natural language processing and computer vision. But despite their popularity, the "correct…