Open spadaval opened 6 years ago
For some context, an AST is an abstract representation of some variable, during program execution. It represents all the possible states of the variable, depending on a bunch of constraints. The goal is to, given a set of AST's, identify what that variable could refer to, and what the program is doing overall.
Is it possible to build a neural net, or some other classification system, that can parse an abstract state tree, and decide what type of behavior it's exhibiting?