Weighted Prediction Divergence for Metareasoning
Weighted Prediction Divergence for Metareasoning
This chapter presents a metareasoning system that relies on a prediction performance measurement, and proposes a novel model performance measurement that fulfills this need: weighted prediction divergence. It begins by proposing a metareasoning system coupled with a robust measurement of model-prediction performance to yield a powerful new method of model selection for an agent. After reviewing some of the existing methods used for performance measurement, it presents a new method for characterizing model prediction performance. It then introduces one target domain for the empirical testing of this prediction-performance measurement. Next, it discusses empirical results using this method and concludes with a description of future work required to realize the overall metareasoning system.
Keywords: metareasoning system, prediction performance measurement, model performance, weighted prediction divergence
MIT Press Scholarship Online requires a subscription or purchase to access the full text of books within the service. Public users can however freely search the site and view the abstracts and keywords for each book and chapter.
Please, subscribe or login to access full text content.
If you think you should have access to this title, please contact your librarian.
To troubleshoot, please check our FAQs, and if you can't find the answer there, please contact us.
