| Presentation | 2022-11-18 Topology analysis in the P2P Federated Learning Yoshihiko Ito, Hideya Ochiai, Hiroshi Esaki, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | P2P Federated Learning is a method of distributed machine learning and the cost of model trading becomes too huge in a large-scale topology. The purpose of this study is to seek a topology that solves cost problems and reduces the influence of Non-IID, allows nodes to add and remove. To achieve these goals, I introduce hierarchy topology.Nodes are divided into several groups and models traded within the group. After that, they traded between groups. The experimental results prove that the new method is better than conventional topology and classifying similar data distributions into the same group shows excellent results. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | topology / P2P Federated Learning / Non-IID / grouping |
| Paper # | CAS2022-50,MSS2022-33 |
| Date of Issue | 2022-11-10 (CAS, MSS) |
| Conference Information | |
| Committee | CAS / MSS / IPSJ-AL |
|---|---|
| Conference Date | 2022/11/17(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yoshinobu Maeda(Niigata Univ.) / Atsuo Ozaki(Osaka Inst. of Tech.) / 全 眞嬉(東北大学) |
| Vice Chair | Yasutoshi Aibara(OmniVision) / Shingo Yamaguchi(Yamaguchi Univ.) |
| Secretary | Yasutoshi Aibara(NIT, Toyama college) / Shingo Yamaguchi(Renesas Electronics) / (Hokkaido Univ.) |
| Assistant | Takahide Sato(Univ. of Yamanashi) / Motoi Yamaguchi(TECHNOPRO) / Shinji Shimoda(Sony Semiconductor Solutions) / Shunsuke Koshita(Hachinohe Inst. of Tech.) / Masato Shirai(Shimane Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Circuits and Systems / Technical Committee on Mathematical Systems Science and its Applications / Special Interest Group on Algorithms |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Topology analysis in the P2P Federated Learning |
| Sub Title (in English) | |
| Keyword(1) | topology |
| Keyword(2) | P2P Federated Learning |
| Keyword(3) | Non-IID |
| Keyword(4) | grouping |
| 1st Author's Name | Yoshihiko Ito |
| 1st Author's Affiliation | The University of Tokyo(Univ, Tokyo) |
| 2nd Author's Name | Hideya Ochiai |
| 2nd Author's Affiliation | The University of Tokyo(Univ, Tokyo) |
| 3rd Author's Name | Hiroshi Esaki |
| 3rd Author's Affiliation | The University of Tokyo(Univ, Tokyo) |
| Date | 2022-11-18 |
| Paper # | CAS2022-50,MSS2022-33 |
| Volume (vol) | vol.122 |
| Number (no) | CAS-253,MSS-254 |
| Page | pp.pp.63-66(CAS), pp.63-66(MSS), |
| #Pages | 4 |
| Date of Issue | 2022-11-10 (CAS, MSS) |