Ask HN: Are there any attempts to employ NN to high level disassembly?

There is a lot solution about code generation by neural networks, but I don't see any attempts on recreating code based on machine code/assembler.

Use cases for it:

1. Transpilation - you have a project in one "old" programming language and you want to smoothly migrate to "new" one

2. Open source licenses sometimes are not followed by vendors, they take open source code, but they don't contribute back. It would allow to at least have a glimpse of what is going on in such piece of code

3. Old, or dropped or proprietary code without access or even existence of source code could be investigated easier

...

...