Document Type
Technical Report
Publication Date
8-1-2014
Technical Report Number
TR2014-763
Abstract
We prove an upper bound on the cumulative opportunity cost of the online knowledge gradient algorithm. We leverage the theory of martingales to yield a bound under the Gaussian assumption. Using results from information theory we are further able to provide asymptotic bounds on the cumulative opportunity cost with high probability.
Dartmouth Digital Commons Citation
Brofos, James, "On Opportunity Cost Bounds for the Knowledge Gradient" (2014). Computer Science Technical Report TR2014-763. https://digitalcommons.dartmouth.edu/cs_tr/364