Grammar Guided Genetic Programming

Grammar-guided genetic programming (GGGP) is an automated machine learning (AutoML) technique that uses evolutionary algorithms to design and optimize machine learning workflows, including algorithm selection and hyperparameter tuning, by representing them as programs generated from a formal grammar. Current research focuses on improving the efficiency and effectiveness of GGGP through interactive optimization, incorporating domain-specific operators, enhancing population diversity, and refining grammar design and initialization methods. This approach holds significant promise for accelerating scientific discovery and practical applications by automating complex data analysis pipelines across diverse fields, such as drug discovery, software engineering, and the solution of partial differential equations.