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

The need for algorithms able to solve Reinforcement Learning (RL) problemswith few trials has motivated the advent of model-based RL methods. Thereported performance of model-based algorithms has dramatically increasedwithin recent years. However, it is not clear how much of the recent progressis due to improved algorithms or due to improved models. While differentmodeling options are available to choose from when applying a model-basedapproach, the distinguishing traits and particular strengths of differentmodels are not clear. The main contribution of this work lies precisely inassessing the model influence on the performance of RL algorithms. A set ofcommonly adopted models is established for the purpose of model comparison.These include Neural Networks (NNs), ensembles of NNs, two differentapproximations of Bayesian NNs (BNNs), that is, the Concrete Dropout NN and theAnchored Ensembling, and Gaussian Processes (GPs). The model comparison isevaluated on a suite of continuous control benchmarking tasks. Our resultsreveal that significant differences in model performance do exist. The ConcreteDropout NN reports persistently superior performance. We summarize thesedifferences for the benefit of the modeler and suggest that the model choice istailored to the standards required by each specific application.

Which Model to Trust: Assessing the Influence of Models on the Performance of Reinforcement Learning Algorithms for Continuous Control Tasks