tianzhaotju/LEAM

We propose a novel DL-based mutation technique (LEAM), which adapts the syntax-guided encoder-decoder architecture to build two sub-models based on AST information, for better learning to represent real faults and ensuring syntactic correctness of mutation faults.