Learning a Behavior Model of Hybrid Systems Through Combining Model-Based Testing and Machine Learning (Full Version)
This work addresses the problem of learning hybrid behavior models for cyber-physical systems, which is incremental as it improves data generation efficiency.
The paper tackles the challenge of constructing models for cyber-physical systems by combining automata learning and model-based testing to automatically generate training data for machine learning, resulting in recurrent neural networks that reduce classification error for crash detection by a factor of five and achieve similar F1-scores with up to three orders of magnitude fewer training samples.
Models play an essential role in the design process of cyber-physical systems. They form the basis for simulation and analysis and help in identifying design problems as early as possible. However, the construction of models that comprise physical and digital behavior is challenging. Therefore, there is considerable interest in learning such hybrid behavior by means of machine learning which requires sufficient and representative training data covering the behavior of the physical system adequately. In this work, we exploit a combination of automata learning and model-based testing to generate sufficient training data fully automatically. Experimental results on a platooning scenario show that recurrent neural networks learned with this data achieved significantly better results compared to models learned from randomly generated data. In particular, the classification error for crash detection is reduced by a factor of five and a similar F1-score is obtained with up to three orders of magnitude fewer training samples.