Hacker News new | past | comments | ask | show | jobs | submit login

By generating the "code" are you merely inferring / representing UI layout as some kind of a tree. E.g. each button/panel/textbox gets detected, and an "RNN/LSTM" generate a tree structure using attention.

"merely" is not the word I would use.

If you somewhat familiar with DL literature, you will see this paper, while having a very interesting angle, the underlying architecture is a standard, enc-dec network, with encoder being CNN and decoding being LSTM. Such application, has been studied before:


The above paper shows nice result that turns image to latex expression, and image to html.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact