Do not Match, Inherit: Fitness Surrogates for Genetics-Based Machine Learning Techniques

The document discusses the development of fitness surrogates for genetics-based machine learning techniques, focusing on enhanced execution efficiency in rule matching and fitness evaluation. It presents an overview of the extended compact classifier system (â‡eccs) using a probabilistic model for accurate rule learning and introduces fitness inheritance methods using least squares fitting. The conclusions emphasize the importance of efficient implementations to reduce the need for matching and identify necessary overlapping building blocks for specific problem types.