Visual Question Answering with Memory-Augmented Networks

Visual Question Answering with Memory-Augmented Networks 2018-05-15 20:15:03 Motivation: 雖然 VQA 已經取得了很大的進步,但是這種方法依然對完全 general,freeform VQA 表現很差,作者認爲是因爲如下兩點:   1. deep models trained with gradient bas
相關文章
相關標籤/搜索