AIMS AND SCOPE
Lexicase selection selects individuals on the basis of their performance on training cases that are considered one at a time, in random order. First developed for parent selection in genetic programming, it has since been applied in other settings both within evolutionary computation and in other forms of machine learning [1,2].
Lexicase selection has been used, for example, for software synthesis, evolutionary robotics, multiobjective optimization, reinforcement learning, learning classifier systems (with a variant called batch lexicase selection), symbolic regression (with a variant called epsilon lexicase selection) and deep learning (with a variant called gradient lexicase selection). It has been analyzed with respect to selection efficiency, population diversity, the promotion of specialists, and problem-solving power, and such analyses have given rise to new methods such as informed downsampled lexicase selection and probabilistic lexicase selection.
This special issue of GPEM will focus on further studies of lexicase selection, its applications, its variants, and its analysis. It aims to draw together ideas from various perspectives to give a more holistic view of lexicase selection, helping guide future research and practice, deepening our understanding of how and why it works and when it should and shouldn't be used.
[1] T. Helmuth, L. Spector and J. Matheson, "Solving Uncompromising Problems With Lexicase Selection," in IEEE Transactions on Evolutionary Computation, vol. 19, no. 5, pp. 630-643, Oct. 2015, doi: 10.1109/TEVC.2014.2362729
[2] Spector, L., Ding, L., Boldi, R. (2024). Particularity. In: Winkler, S., Trujillo, L., Ofria, C., Hu, T. (eds) Genetic Programming Theory and Practice XX. Genetic and Evolutionary Computation. Springer, Singapore. https://doi.org/10.1007/978-981-99-8413-8_9
TOPICS OF INTEREST
- Analysis of the benefits and drawbacks of lexicase selection
- Analysis of applications that are amenable (or not) to lexicase selection
- Studies of variants of lexicase selection
- Creation of new variants of lexicase selection
- Critiques or position papers considering lexicase selection
- The lexicase algorithm used beyond parent selection
- Lexicase selection in contexts beyond genetic programming, including but not limited to other evolutionary algorithms, evolutionary robotics, reinforcement learning, and neural networks
IMPORTANT DATES
- Submission deadline: November 30, 2026
- Notification of first review: January 15, 2027
- Submission of revised manuscript: March 15, 2027
- Notification of final decision: April 15, 202
GUEST EDITORS
- Thomas Helmuth, Hamilton College, Clinton, United States, thelmuth@hamilton.edu
- Lee Spector, Swoop Inc., United States, lspector@amherst.edu
- Edward Pantridge, Amherst College, Amherst Center, United States, eddie.pantridge@gmail.com
