| Presentation | 2021-12-17 [Short Paper] Study on Improving the Characteristics of Random Walk on Graph using Q-learning Tomoyuki Miyashita, Taisei Suzuki, Ryotaro Matsuo, Hiroyuki Ohsaki, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In recent years, modeling mobile agent on unknown graphs, such as random walks on graphs and understanding its mathematical properties have been studied. The mobility models of agents on graphs has also began to be applied to network exploration and information search on networks. It is not easy to improve the properties of the mobility models on graphs, because the information available to the mobile agents is very limited. In this paper, we investigate to what extent the properties of random walks can be improved when the mobile agents has access to very limited information. In particular, through experiments, we examine how much the properties of random walk can be improved using a kind of machine learning, reinforcement learning. Specifically, we propose a random walk based on Q-learning (QW-RW; Q-Weighted Random Walk), in which an agent decides a destination node using Q-values learned by Q-learning, one of the reinforcement learning techniques. Furthermore, through simulation experiments, we examine the effectiveness of the QW-RW. Our findings include that the QW-RW mobile agent covered the graph as fast as or slightly faster than the typical mobile model based on a random walk. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Q-Weighted Random Walk / Random Walk / Q-learning / Mobility Model / Reinforcement Learning |
| Paper # | IA2021-51 |
| Date of Issue | 2021-12-09 (IA) |
| Conference Information | |
| Committee | IN / IA |
|---|---|
| Conference Date | 2021/12/16(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Higashi-Senda campus, Hiroshima Univ. |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Performance Analysis and Simulation, Robustness, Traffic and Throughput Measurement, Quality of Service (QoS) Control, Congestion Control, Overlay Network/P2P, IPv6, Multicast, Routing, DDoS, etc. |
| Chair | Kenji Ishida(Hiroshima City Univ.) / Tomoki Yoshihisa(Osaka Univ.) |
| Vice Chair | Kunio Hato(Internet Multifeed) / Toru Kondo(Hiroshima Univ.) / Yuichiro Hei(KDDI Research) / Hiroshi Yamamoto(Ritsumeikan Univ.) |
| Secretary | Kunio Hato(NTT) / Toru Kondo(Univ. of Nagasaki) / Yuichiro Hei(Nagaoka Univ. of Tech.) / Hiroshi Yamamoto(KDDI Research) |
| Assistant | / Daisuke Kotani(Kyoto Univ.) / Ryo Nakamurai(Fukuoka Univ.) / Daiki Nobayashi(Kyushu Inst. of Tech.) |
| Paper Information | |
| Registration To | Technical Committee on Information Networks / Technical Committee on Internet Architecture |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | [Short Paper] Study on Improving the Characteristics of Random Walk on Graph using Q-learning |
| Sub Title (in English) | |
| Keyword(1) | Q-Weighted Random Walk |
| Keyword(2) | Random Walk |
| Keyword(3) | Q-learning |
| Keyword(4) | Mobility Model |
| Keyword(5) | Reinforcement Learning |
| 1st Author's Name | Tomoyuki Miyashita |
| 1st Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
| 2nd Author's Name | Taisei Suzuki |
| 2nd Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
| 3rd Author's Name | Ryotaro Matsuo |
| 3rd Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
| 4th Author's Name | Hiroyuki Ohsaki |
| 4th Author's Affiliation | Kwansei Gakuin University(Kwansei Gakuin Univ.) |
| Date | 2021-12-17 |
| Paper # | IA2021-51 |
| Volume (vol) | vol.121 |
| Number (no) | IA-300 |
| Page | pp.pp.100-103(IA), |
| #Pages | 4 |
| Date of Issue | 2021-12-09 (IA) |