Depth over Fidelity in Fixed-Budget Noisy Evolution Strategies
Researchers have developed a new method called Probabilistic Elite Membership (PEM) to improve noisy evolution strategies under fixed evaluation budgets. This approach prioritizes exploring more distribution updates (depth) over refining the accuracy of each update (fidelity). PEM integrates ranking uncertainty to maintain the conditional mean update while reducing dispersion, effectively optimizing policy search and hyperparameter tuning in budget-constrained scenarios. AI
IMPACT Enhances optimization techniques for AI tasks like policy search and hyperparameter tuning.