%0 Conference Paper %1 YaoSze12 %A Yao, H. %A Szepesvári, Cs. %B AAAI-2012 %D 2012 %K Decision LSTD Markov Processes,function approximation, control control, difference learning, planning, reinforcement temporal %P 1212--1217 %T Approximate Policy Iteration with Linear Action Models %X In this paper we consider the problem of finding a good policy given some batch data. We propose a new approach, LAM-API, that first builds a so-called linear action model (LAM) from the data and then uses the learned model and the collected data in approximate policy iteration (API) to find a good policy. A natural choice for the policy evaluation step in this algorithm is to use least-squares temporal difference (LSTD) learning algorithm. Empirical results on three benchmark problems show that this particular instance of LAM-API performs competitively as compared with LSPI, both from the point of view of data and computational efficiency.

@inproceedings{YaoSze12, abstract = {In this paper we consider the problem of finding a good policy given some batch data. We propose a new approach, LAM-API, that first builds a so-called linear action model (LAM) from the data and then uses the learned model and the collected data in approximate policy iteration (API) to find a good policy. A natural choice for the policy evaluation step in this algorithm is to use least-squares temporal difference (LSTD) learning algorithm. Empirical results on three benchmark problems show that this particular instance of LAM-API performs competitively as compared with LSPI, both from the point of view of data and computational efficiency.}, added-at = {2020-03-17T03:03:01.000+0100}, author = {Yao, H. and Szepesv{\'a}ri, {Cs}.}, biburl = {https://www.bibsonomy.org/bibtex/2d01e79cab0ea85b4a413d329b67fe133/csaba}, booktitle = {AAAI-2012}, date-added = {2012-06-06 14:33:03 -0600}, date-modified = {2013-07-16 12:05:29 -0600}, interhash = {4a13e85f513c9fd3ce4fb12d1e7d7034}, intrahash = {d01e79cab0ea85b4a413d329b67fe133}, keywords = {Decision LSTD Markov Processes,function approximation, control control, difference learning, planning, reinforcement temporal}, month = {July}, pages = {1212--1217}, pdf = {papers/lamapi.pdf}, timestamp = {2020-03-17T03:03:01.000+0100}, title = {Approximate Policy Iteration with Linear Action Models}, year = 2012 }