Lexicase Selection

Lexicase selection is a parent selection algorithm used in evolutionary computation, particularly genetic programming and automated machine learning, that prioritizes individuals based on their performance across individual test cases rather than aggregated metrics. Current research focuses on improving its efficiency (e.g., through diverse aggregation and weighted shuffling) and effectiveness, especially when dealing with large datasets or conflicting objectives, often incorporating techniques like down-sampling and epsilon-thresholds. These advancements enhance the algorithm's applicability to complex problems in various fields, including symbolic regression, deep learning, and recommender systems, by improving both solution quality and computational speed.