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

This paper proposes a new algorithm, referred to as GMAB, that combinesconcepts from the reinforcement learning domain of multi-armed bandits andrandom search strategies from the domain of genetic algorithms to solvediscrete stochastic optimization problems via simulation. In particular, thefocus is on noisy large-scale problems, which often involve a multitude ofdimensions as well as multiple local optima. Our aim is to combine the propertyof multi-armed bandits to cope with volatile simulation observations with theability of genetic algorithms to handle high-dimensional solution spacesaccompanied by an enormous number of feasible solutions. For this purpose, amulti-armed bandit framework serves as a foundation, where each observedsimulation is incorporated into the memory of GMAB. Based on this memory,genetic operators guide the search, as they provide powerful tools forexploration as well as exploitation. The empirical results demonstrate thatGMAB achieves superior performance compared to benchmark algorithms from theliterature in a large variety of test problems. In all experiments, GMABrequired considerably fewer simulations to achieve similar or (far) bettersolutions than those generated by existing methods. At the same time, GMAB'soverhead with regard to the required runtime is extremely small due to thesuggested tree-based implementation of its memory. Furthermore, we prove itsconvergence to the set of global optima as the simulation effort goes toinfinity.

Genetic multi-armed bandits: a reinforcement learning approach for discrete optimization via simulation