Introducing XCS to Coverage Directed test Generation

From OpenAlex, which stores abstracts as an index of words and their positions: what is shown is rebuilt from that, and reads fluently whether or not it is what the author wrote.

Abstract

Coverage Directed test Generation (CDG) is rife with challenges and problems, despite the relative successes of machine learning methodologies over the years in automating it. This paper introduces the use of the eXtended Classifier System (XCS) in simulation-based digital design verification. It argues for the use of this novel genetics-based machine learning technique to perform effective CDG by learning the full mapping between coverage results and test generator directives. Using the resulting production rules, efficient test suites can be constructed, and inference on the validity of the verification environment can be made. There is great potential in using XCS for design verification and this paper forms an initial attempt to highlight the associated advantages. The technique requires no domain knowledge to setup and satisfies important CDG requirements. Once matured, it is expected to be utilized seamlessly in any industrial level simulation-based verification process.