Hyper-optimization with Gaussian Process and Differential Evolution Algorithm
arXiv:2101.10625
Abstract
Optimization of problems with high computational power demands is a challenging task. A probabilistic approach to such optimization called Bayesian optimization lowers performance demands by solving mathematically simpler model of the problem. Selected approach, Gaussian Process, models problem using a mixture of Gaussian functions. This paper presents specific modifications of Gaussian Process optimization components from available scientific libraries. Presented modifications were submitted to BlackBox 2020 challenge, where it outperformed some conventionally available optimization libraries.
BlackBox2020 challenge track
References in corpus (4)
- Practical Bayesian Optimization of Machine Learning Algorithms
- A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay
- Dealing with Integer-valued Variables in Bayesian Optimization with Gaussian Processes
- Dealing with Categorical and Integer-valued Variables in Bayesian Optimization with Gaussian Processes