SLGPT: Using Transfer Learning to Directly Generate Simulink Model Files and Find Bugs in the Simulink Toolchain
This addresses bug detection in commercial CPS tools like Simulink, which is challenging due to large codebases and lack of formal specifications, representing an incremental improvement over existing methods.
The paper tackled the problem of finding bugs in Simulink, a complex cyber-physical system tool, by using transfer learning with GPT-2 to generate Simulink models directly; SLGPT produced models more similar to open-source ones than DeepFuzzSL and found a super-set of bugs.
Finding bugs in a commercial cyber-physical system (CPS) development tool such as Simulink is hard as its codebase contains millions of lines of code and complete formal language specifications are not available. While deep learning techniques promise to learn such language specifications from sample models, deep learning needs a large number of training data to work well. SLGPT addresses this problem by using transfer learning to leverage the powerful Generative Pre-trained Transformer 2 (GPT-2) model, which has been pre-trained on a large set of training data. SLGPT adapts GPT-2 to Simulink with both randomly generated models and models mined from open-source repositories. SLGPT produced Simulink models that are both more similar to open-source models than its closest competitor, DeepFuzzSL, and found a super-set of the Simulink development toolchain bugs found by DeepFuzzSL.