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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Optimization algorithms are very different from human optimizers. A humanbeing would gain more experiences through problem-solving, which helps her/himin solving a new unseen problem. Yet an optimization algorithm never gains anyexperiences by solving more problems. In recent years, efforts have been madetowards endowing optimization algorithms with some abilities of experiencelearning, which is regarded as experience-based optimization. In this paper, weargue that hard optimization problems could be tackled efficiently by makingbetter use of experiences gained in related problems. We demonstrate our ideasin the context of expensive optimization, where we aim to find a near-optimalsolution to an expensive optimization problem with as few fitness evaluationsas possible. To achieve this, we propose an experience-based surrogate-assistedevolutionary algorithm (SAEA) framework to enhance the optimization efficiencyof expensive problems, where experiences are gained across related expensivetasks via a novel meta-learning method. These experiences serve as thetask-independent parameters of a deep kernel learning surrogate, then thesolutions sampled from the target task are used to adapt task-specificparameters for the surrogate. With the help of experience learning, competitiveregression-based surrogates can be initialized using only 1$d$ solutions fromthe target task ($d$ is the dimension of the decision space). Our experimentalresults on expensive multi-objective and constrained optimization problemsdemonstrate that experiences gained from related tasks are beneficial for thesaving of evaluation budgets on the target problem.

Experience-Based Evolutionary Algorithms for Expensive Optimization