English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Bayesian optimization (BO) has demonstrated potential for optimizing controlperformance in data-limited settings, especially for systems with unknowndynamics or unmodeled performance objectives. The BO algorithm efficientlytrades-off exploration and exploitation by leveraging uncertainty estimatesusing surrogate models. These surrogates are usually learned using datacollected from the target dynamical system to be optimized. Intuitively, theconvergence rate of BO is better for surrogate models that can accuratelypredict the target system performance. In classical BO, initial surrogatemodels are constructed using very limited data points, and therefore rarelyyield accurate predictions of system performance. In this paper, we propose theuse of meta-learning to generate an initial surrogate model based on datacollected from performance optimization tasks performed on a variety of systemsthat are different to the target system. To this end, we employ deep kernelnetworks (DKNs) which are simple to train and which comprise encoded Gaussianprocess models that integrate seamlessly with classical BO. The effectivenessof our proposed DKN-BO approach for speeding up control system performanceoptimization is demonstrated using a well-studied nonlinear system with unknowndynamics and an unmodeled performance function.

Optimizing Closed-Loop Performance with Data from Similar Systems: A Bayesian Meta-Learning Approach