| Presentation | 2011-06-24 Solving POMDPs using Restricted Boltzmann Machines with Echo State Networks Makoto OTSUKA, Junichiro YOSHIMOTO, Stefan ELFWING, Kenji DOYA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | A partially observable Markov decision process (POMDP) can be solved in a model-based way using explicit knowledge of the environmental dynamics or in a model-free way using implicit representations of task-relevant states. Here we consider a model-free approach of combining an echo state network (ESN) for summarizing past actions and observations and a restricted Boltzmann machine (RBM) for learning action values in a high-dimensional state space. Simulation results in robot navigation tasks showed that the ESN can capture relevant information in the sequence of high dimensional observations and that RBM can construct task-oriented internal representation in its hidden layer. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | partially observable Markov decision processes / restricted Boltzmann machines / echo state networks / goal-directed representation / free energy |
| Paper # | NC2011-19 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2011/6/16(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Neurocomputing (NC) |
|---|---|
| Language | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Solving POMDPs using Restricted Boltzmann Machines with Echo State Networks |
| Sub Title (in English) | |
| Keyword(1) | partially observable Markov decision processes |
| Keyword(2) | restricted Boltzmann machines |
| Keyword(3) | echo state networks |
| Keyword(4) | goal-directed representation |
| Keyword(5) | free energy |
| 1st Author's Name | Makoto OTSUKA |
| 1st Author's Affiliation | Neural Computation Unit, Okinawa Institute of Science and Technology() |
| 2nd Author's Name | Junichiro YOSHIMOTO |
| 2nd Author's Affiliation | Neural Computation Unit, Okinawa Institute of Science and Technology |
| 3rd Author's Name | Stefan ELFWING |
| 3rd Author's Affiliation | Neural Computation Unit, Okinawa Institute of Science and Technology |
| 4th Author's Name | Kenji DOYA |
| 4th Author's Affiliation | Neural Computation Unit, Okinawa Institute of Science and Technology |
| Date | 2011-06-24 |
| Paper # | NC2011-19 |
| Volume (vol) | vol.111 |
| Number (no) | 96 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |