%0 Journal Article %1 JiPrFuMaSz18 %A Jie, C. %A Prashanth L.A., %A Fu, M.C. %A Marcus, S. %A Szepesvári, Cs. %D 2018 %J IEEE Transactions on Automatic Control %K Decision Markov Processes, SPSA, criteria, cumulative learning, optimization, prospect reinforcement risk-sensitive stochastic theory %N 9 %P 2867--2882 %T Stochastic Optimization in a Cumulative Prospect Theory Framework %U http://ieeexplore.ieee.org/document/8014469/ %V 63 %X Cumulative prospect theory (CPT) is a popular approach for modeling human preferences. It is based on probabilistic distortions and generalizes expected utility theory. We bring CPT to a stochastic optimization framework and propose algorithms for both estimation and optimization of CPT-value objectives. We propose an empirical distribution function-based scheme to estimate the CPT-value and then use this scheme in the inner loop of a CPT-value optimization procedure. We propose both gradient-based as well as gradient-free CPT-value optimization algorithms that are based on two well-known simulation optimization ideas: simultaneous perturbation stochastic approximation (SPSA) and model-based parameter search (MPS), respectively. We provide theoretical convergence guarantees for all the proposed algorithms and also illustrate the potential of CPT-based criteria in a traffic signal control application.

@article{JiPrFuMaSz18, abstract = {Cumulative prospect theory (CPT) is a popular approach for modeling human preferences. It is based on probabilistic distortions and generalizes expected utility theory. We bring CPT to a stochastic optimization framework and propose algorithms for both estimation and optimization of CPT-value objectives. We propose an empirical distribution function-based scheme to estimate the CPT-value and then use this scheme in the inner loop of a CPT-value optimization procedure. We propose both gradient-based as well as gradient-free CPT-value optimization algorithms that are based on two well-known simulation optimization ideas: simultaneous perturbation stochastic approximation (SPSA) and model-based parameter search (MPS), respectively. We provide theoretical convergence guarantees for all the proposed algorithms and also illustrate the potential of CPT-based criteria in a traffic signal control application. }, added-at = {2020-03-17T03:03:01.000+0100}, author = {Jie, C. and {Prashanth L.A.} and Fu, M.C. and Marcus, S. and {Sz}epesv{\'a}ri, {Cs}.}, bdsk-file-1 = {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}, bdsk-url-1 = {http://ieeexplore.ieee.org/document/8014469/}, bdsk-url-2 = {https://dx.doi.org/10.1109/TAC.2017.2743163}, biburl = {https://www.bibsonomy.org/bibtex/2a4a65a3d2e7627184cd06bdbdf801676/csaba}, date-added = {2018-03-11 17:24:12 +0000}, date-modified = {2019-07-20 10:15:20 -0600}, interhash = {c6028afaff1ae3b44a92d2ac48a64558}, intrahash = {a4a65a3d2e7627184cd06bdbdf801676}, journal = {IEEE Transactions on Automatic Control}, keywords = {Decision Markov Processes, SPSA, criteria, cumulative learning, optimization, prospect reinforcement risk-sensitive stochastic theory}, number = 9, pages = {2867--2882}, pdf = {papers/2018-cpt-rl-tac.pdf}, rating = {0}, read = {Yes}, timestamp = {2020-03-17T03:03:01.000+0100}, title = {Stochastic Optimization in a Cumulative Prospect Theory Framework}, url = {http://ieeexplore.ieee.org/document/8014469/}, volume = 63, year = 2018 }