C. Dimitrakakis, and C. Rothkopf. (2011)cite arxiv:1106.3655Comment: Corrected version. 13 pages, 8 figures.
Abstract
We generalise the problem of inverse reinforcement learning to multiple
tasks, from multiple demonstrations. Each one may represent one expert trying
to solve a different task, or as different experts trying to solve the same
task. Our main contribution is to formalise the problem as statistical
preference elicitation, via a number of structured priors, whose form captures
our biases about the relatedness of different tasks or expert policies. In
doing so, we introduce a prior on policy optimality, which is more natural to
specify. We show that our framework allows us not only to learn to efficiently
from multiple experts but to also effectively differentiate between the goals
of each. Possible applications include analysing the intrinsic motivations of
subjects in behavioural experiments and learning from multiple teachers.
%0 Generic
%1 dimitrakakis2011bayesian
%A Dimitrakakis, Christos
%A Rothkopf, Constantin
%D 2011
%K bayesian inverse learning multitask reinforcement
%T Bayesian multitask inverse reinforcement learning
%U http://arxiv.org/abs/1106.3655
%X We generalise the problem of inverse reinforcement learning to multiple
tasks, from multiple demonstrations. Each one may represent one expert trying
to solve a different task, or as different experts trying to solve the same
task. Our main contribution is to formalise the problem as statistical
preference elicitation, via a number of structured priors, whose form captures
our biases about the relatedness of different tasks or expert policies. In
doing so, we introduce a prior on policy optimality, which is more natural to
specify. We show that our framework allows us not only to learn to efficiently
from multiple experts but to also effectively differentiate between the goals
of each. Possible applications include analysing the intrinsic motivations of
subjects in behavioural experiments and learning from multiple teachers.
@misc{dimitrakakis2011bayesian,
abstract = {We generalise the problem of inverse reinforcement learning to multiple
tasks, from multiple demonstrations. Each one may represent one expert trying
to solve a different task, or as different experts trying to solve the same
task. Our main contribution is to formalise the problem as statistical
preference elicitation, via a number of structured priors, whose form captures
our biases about the relatedness of different tasks or expert policies. In
doing so, we introduce a prior on policy optimality, which is more natural to
specify. We show that our framework allows us not only to learn to efficiently
from multiple experts but to also effectively differentiate between the goals
of each. Possible applications include analysing the intrinsic motivations of
subjects in behavioural experiments and learning from multiple teachers.},
added-at = {2012-05-27T11:57:22.000+0200},
author = {Dimitrakakis, Christos and Rothkopf, Constantin},
biburl = {https://www.bibsonomy.org/bibtex/26322b1d11ee3acc73300747d9513ca0a/olethros},
description = {Bayesian multitask inverse reinforcement learning},
interhash = {ea5de7bf76b4badfbbc855ff9393c187},
intrahash = {6322b1d11ee3acc73300747d9513ca0a},
keywords = {bayesian inverse learning multitask reinforcement},
note = {cite arxiv:1106.3655Comment: Corrected version. 13 pages, 8 figures},
timestamp = {2012-05-27T11:57:22.000+0200},
title = {Bayesian multitask inverse reinforcement learning},
url = {http://arxiv.org/abs/1106.3655},
year = 2011
}