New GAME Algorithm Illuminates Both Sides Of Adversarial Problems
We applied it to a variety of tasks, such as Parabellum, our battle game research environment. This video shows 2D PCA of the VEM behavior space after running GAME in Parabellum for 2M evaluations over 20 generations.
Sebastian Risi@risi1979We're excited to announce GAME: Adversarial Coevolutionary Illumination with Generational Adversarial MAP-Elites ⚔️ Game is a new coevolutionary QD algorithm that illuminates both sides of an adversarial problem by alternating the evolution of solutions on one side that maximize the adversarial fitness against fixed opponents from the other side. If you have any tasks requiring adversarial training, check it out! Blog: https://game-approach.github.io/ Paper: https://arxiv.org/pdf/2505.06617v3
Or wrestling, which is implemented as a custom EvoGym environment, in which two 2D soft robots fight to be the closest to the center of the arena.
Sebastian Risi@risi1979We applied it to a variety of tasks, such as Parabellum, our battle game research environment. This video shows 2D PCA of the VEM behavior space after running GAME in Parabellum for 2M evaluations over 20 generations.