Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Inverse Optimization for Policy Identification



Chithrupa Ramesh, Marius Schmitt, Peyman Mohajerin Esfahani

In data-driven inverse optimization an observer aims to learn the preferences of an agent who solves a parametric optimization problem depending on an exogenous signal. We are interested in the visual search problem, where agents identify the location of a target symbol in an N×N visual array. By observing actions of expert agents, we wish to use inverse optimisation to learn the underlying cost function for this problem. We use an existing visual search policy in lieu of an expert agent, to generate observations for the inverse-­optimization problem. Our goal is to compare the estimated decisions with the decisions taken using the policy.

Weitere Informationen

John Lygeros

Art der Arbeit:
Voraussetzungen: Optimisation Theory and Basic Control Theory, some knowledge of Markov processes will be useful.
Anzahl StudentInnen:
Status: taken
Projektstart: July 1st, 2016