Faster Convergence with Lexicase Selection in Tree-based Automated Machine Learning
This work addresses convergence speed for users of automated machine learning systems, but it is incremental as it compares existing selection methods without introducing a new paradigm.
The study tackled the problem of slow convergence in evolutionary automated machine learning by comparing parent selection methods, finding that lexicase selection leads to significantly faster convergence compared to NSGA-II in the Tree-based Pipeline Optimization Tool (TPOT) across multiple datasets.
In many evolutionary computation systems, parent selection methods can affect, among other things, convergence to a solution. In this paper, we present a study comparing the role of two commonly used parent selection methods in evolving machine learning pipelines in an automated machine learning system called Tree-based Pipeline Optimization Tool (TPOT). Specifically, we demonstrate, using experiments on multiple datasets, that lexicase selection leads to significantly faster convergence as compared to NSGA-II in TPOT. We also compare the exploration of parts of the search space by these selection methods using a trie data structure that contains information about the pipelines explored in a particular run.