InformationTheoretic Bounded Rationality
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Zoubin Ghahramani.
In this talk I provide an overview of informationtheoretic bounded rationality for planning in sequential decision problems. I show how to ground the theory on a stochastic computation model for largescale choice spaces and then derive the free energy functional as the associated variational principle for characterising boundedrational decisions. These decision processes have three important properties: they trade off utility and decision complexity; they give rise to an equivalence class of behaviourally indistinguishable decision problems; and they possess natural stochastic choice algorithms. I will discuss a general class of boundedrational sequential planning problems that encompasses some wellknown classical planning algorithms as limit cases (such as Expectimax and Minimax), as well as trust and risksensitive planning. Finally, I will point out formal connections to Bayesian inference and to regret theory.
Credits: This is joint work with Daniel A. Braun, KeeEung Kim, Daniel D. Lee, and Naftali Tishby (in alphabetical order).
This talk is part of the Machine Learning @ CUED series.
This talk is included in these lists:
Note that exdirectory lists are not shown.
