NeFut Logo NeFut
Admin Login

[CS.AI] Neuroevolution Arena: Nested Ecological Evaluation of Update-and-Inheritance Regimes across Neural Architectures

Published at: 2026-08-12 22:00 Last updated: 2026-08-13 01:53
#LLM #Neural #Artificial Intelligence

This paper presents Neuroevolution Arena, a GPU-accelerated spatial ecology for evaluating independently parameterized neural-network cells, using an audit-tracked nested evaluation protocol. Three implementation-specific update-and-inheritance regimes (EvoEvo, EvoRL, and RLRL) are crossed with two neural architectures for 50,000 generations in three independent training runs per condition. One saved elite-controller artifact from each of the 18 runs enters an aligned-run frozen-evaluation design comprising 198 computational jobs. The results show that RL-enabled regimes attain higher recorded training fitness than EvoEvo, whereas pairwise outcomes exhibit architecture-conditioned majority patterns and substantial artifact dependence. Six-way winners vary across artifacts and contexts, and the prespecified survival endpoint has a complete floor. We contribute a nested protocol that separates training-run artifacts from evaluation contexts and exposes, rather than conceals, their different sources of variation. Blogger's Review: This paper proposes an innovative neuroevolution evaluation framework that can deeply investigate the performance of different update and inheritance mechanisms in neural architectures, with significant theoretical and practical implications.

Original Source: https://arxiv.org/abs/2608.10323

[h] Back to Home