| Presentation | 2020-12-14 A Study on Relaxation of Network Restriction in Deep-Unfolding aided Consensus Shoya Ogawa, Ishii Koji, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In the consensus problem with a complex network, the convergence performance deeply depends on the given parameters, i.e., the weighting values given at individual edges. Recently, Kishida et.al., have proposed to apply a deep learning technique into the consensus problem and shown that the deep-leaning aided consensus problem can significantly enhance the convergence performance. The deep-learning aided consensus problem tries to learn the samples of the consensus problem with the fixed network topology but different initial values. However, since this system is designed only for the given network topology, Kishida’s deep-learning aided consensus cannot apply to the system with other network topology. To relax the restriction about applying network topology, this study proposes a statistical approach to give the wighting values at individual edges. We first gather the weight values calculated by Kishida’s deep-learning aided consensus with different network topologies and sort the gathered data into the cases corresponding to the time and the number of neighbor agents. From the computer simulation, the shape of pdf of sorted data can be seen as a Gaussian distribution and thus we set up a hypothesis that optimal weighting value follows a Gaussian distribution. Then, this study proposes to assign the random variable which follows the Gaussian distribution into the edge with the same time-index and number of neighbors. Computer simulations show that the proposed system cannot achieve better performance than the conventional system with fixed weighting value. Thus, this work tries to investigate the causes why the proposed deep-learning aided consensus cannot efficiently work. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | consensus problem / data-driven algorithm / deep-unfolding / gaussian approximation |
| Paper # | WBS2020-11,ITS2020-7,RCC2020-14 |
| Date of Issue | 2020-12-07 (WBS, ITS, RCC) |
| Conference Information | |
| Committee | ITS / WBS / RCC |
|---|---|
| Conference Date | 2020/12/14(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Online |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | ITS Communications, Reliable Communication and Control, Radar and Sensing, etc. |
| Chair | Tomotaka Wada(Kansai Univ.) / Masanori Hamamura(Kochi Univ. of Tech.) / HUAN-BANG LI(NICT) |
| Vice Chair | Yusuke Takatori(Kanagawa Inst. of Tech.) / Hiroyuki Hatano(Mie Univ.) / Takashi Shono(INTEL) / Masahiro Fujii(Utsunomiya Univ.) / Shunichi Azuma(Nagoya Univ.) / Koji Ishii(Kagawa Univ.) |
| Secretary | Yusuke Takatori(Univ. of Tokyo) / Hiroyuki Hatano(Akita Prefectural Univ.) / Takashi Shono(Okayama Univ. of Science) / Masahiro Fujii(National Defence Academy) / Shunichi Azuma(CRIEPI) / Koji Ishii(Osaka Univ.) |
| Assistant | Msataka Imao(Mitsubishi Electric) / Yanlei Gu(Ritsumeikan Univ.) / Kenshi Saho(Toyama Prefectural Univ.) / Duong Quang Thang(NAIST) / Masafumi Moriyama(NICT) / Masayuki Kinoshita(Chiba Univ. of Tech.) / SHAN LIN(NICT) / Masaki Ogura(Osaka Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Intelligent Transport Systems Technology / Technical Committee on Wideband System / Technical Committee on Reliable Communication and Control |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | A Study on Relaxation of Network Restriction in Deep-Unfolding aided Consensus |
| Sub Title (in English) | |
| Keyword(1) | consensus problem |
| Keyword(2) | data-driven algorithm |
| Keyword(3) | deep-unfolding |
| Keyword(4) | gaussian approximation |
| 1st Author's Name | Shoya Ogawa |
| 1st Author's Affiliation | Kagawa University(Kagawa Univ.) |
| 2nd Author's Name | Ishii Koji |
| 2nd Author's Affiliation | Kagawa University(Kagawa Univ.) |
| Date | 2020-12-14 |
| Paper # | WBS2020-11,ITS2020-7,RCC2020-14 |
| Volume (vol) | vol.120 |
| Number (no) | WBS-290,ITS-291,RCC-292 |
| Page | pp.pp.19-24(WBS), pp.19-24(ITS), pp.19-24(RCC), |
| #Pages | 6 |
| Date of Issue | 2020-12-07 (WBS, ITS, RCC) |