Loading...
Please wait, while we are loading the content...
Similar Documents
Draft: Please do not distribute Learning and Discovery of Predictive State Representations in Dynamical Systems with Reset
| Content Provider | CiteSeerX |
|---|---|
| Author | James, Michael R. Singh, Satinder |
| Abstract | Predictive state representations (PSRs) have been recently proposed as a new way of modeling controlled dynamical systems. PSR-based models use predictions of the observable outcomes of tests that could be done on the system as their state representation, and have model parameters that define how the predictive state representation changes over time as actions are taken and observations noted. Learning PSR-based models requires solving two subproblems: 1) discovery of the tests whose predictions constitute state, and 2) learning the model parameters that define the dynamics. So far, there have been no results available on the discovery subproblem while for the learning subproblem an approximategradient algorithm has been proposed [7] with mixed results (it works on some domains and not on others). In this paper, we provide the first PSR-discovery algorithm and a new learning algorithm for the special class of controlled dynamical systems that have a reset operation. We provide preliminary experimental verification of our algorithms. 1 |
| File Format | |
| Access Restriction | Open |
| Subject Keyword | Predictive State Representation Dynamical System Controlled Dynamical System Psr-based Model Mixed Result State Representation Discovery Subproblem Observable Outcome Predictive State Representation Change New Learning Algorithm Special Class Reset Operation Approximategradient Algorithm Model Parameter Preliminary Experimental Verification First Psr-discovery Algorithm |
| Content Type | Text |
