This PhD dissertation presents a collection of novel, state-of-the-art algorithms for solving problems in the class of continuous black-box optimization. Natural Evolution Strategies are a family of algorithms that constitutes a general-purpose approach. Maintaining a parameterized distribution on the set of solution candidates, the natural gradient is used to update the distribution's parameters in the direction of higher expected fitness. Techniques are introduced that addresses issues of convergence, robustness, computational complexity and algorithm speed. We also demonstrate how the principle of artificial curiosity can guide exploration in the context of costly optimization, introducing a response surface method that estimates the interestingness of each candidate point using Gaussian process regression. The results show best published performance on various standard benchmarks, as well as competitive performance on others.
Lulu Staff has been notified of a possible violation of the terms of our Membership Agreement. Our agents will determine if the content reported is inappropriate or not based on the guidelines provided and will then take action where needed.
Thank you for notifying us. We will email you with the results and/or actions taken as a result of the investigation if you chose to receive confirmation.