| 電子情報通信学会 研究会発表申込システム 講演論文 詳細 |
技報閲覧サービス [ログイン] 技報アーカイブ |
| トップに戻る | 前のページに戻る | [Japanese] / [English] |
| 講演抄録／キーワード | ||
| 講演名 | 2023-08-31 16:20 Enhancing Communication Efficiency for UAV Networks through Knowledge Distillation and Transfer Learning in Federated Learning ○Yalong Li・Zhaoyang Du・Celimuge Wu・Tsutomu Yoshinaga（UEC）  CQ2023-30 |
|
| 抄録 | （和） | Federated learning (FL) in unmanned aerial vehicles (UAVs) networks demands considerable communication resources to transfer model data between the central server and UAVs (FL clients). However, different UAVs may have different communication capabilities due to the UAV's maneuverability and heterogeneity, where limited communication resource could be bottle neck for FL performance. In this paper, we first introduce a knowledge distillation based approach that places two different models with different sizes, namely the teacher model and student model, for FL client to make a trade-off between the FL performance and communication cost. Then, we propose a novel model switching method to switch between the teacher model and student model to adapt to the dynamic feature of UAV networks. Specifically, considering available communication bandwidth and learning accuracy, we design a threshold-based model switching algorithm (TBMSA) and determine the threshold based on the k-means method (DTBKM) to accurately and quickly determine the switching point. In addition, for the knowledge transfer between models, we design a knowledge inheritance based on a transfer learning (KIBTL) algorithm, which transfers knowledge from one model to another. Experiments show that the proposed model switching algorithm achieves significant performance improvements as compared to existing baselines. |
| （英） | Federated learning (FL) in unmanned aerial vehicles (UAVs) networks demands considerable communication resources to transfer model data between the central server and UAVs (FL clients). However, different UAVs may have different communication capabilities due to the UAV's maneuverability and heterogeneity, where limited communication resource could be bottle neck for FL performance. In this paper, we first introduce a knowledge distillation based approach that places two different models with different sizes, namely the teacher model and student model, for FL client to make a trade-off between the FL performance and communication cost. Then, we propose a novel model switching method to switch between the teacher model and student model to adapt to the dynamic feature of UAV networks. Specifically, considering available communication bandwidth and learning accuracy, we design a threshold-based model switching algorithm (TBMSA) and determine the threshold based on the k-means method (DTBKM) to accurately and quickly determine the switching point. In addition, for the knowledge transfer between models, we design a knowledge inheritance based on a transfer learning (KIBTL) algorithm, which transfers knowledge from one model to another. Experiments show that the proposed model switching algorithm achieves significant performance improvements as compared to existing baselines. | |
| キーワード | （和） | UAV Networks / Federated Learning / Model Switching / Knowledge Distillation / Transfer Learning / / / |
| （英） | UAV Networks / Federated Learning / Model Switching / Knowledge Distillation / Transfer Learning / / / | |
| 文献情報 | 信学技報, vol. 123, no. 174, CQ2023-30, pp. 26-31, 2023年8月. | |
| 資料番号 | CQ2023-30 | |
| 発行日 | 2023-08-24 (CQ) | |
| ISSN | Online edition: ISSN 2432-6380 | |
| 著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します．(許諾番号：10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| PDFダウンロード | CQ2023-30 |
|
| 研究会情報 | |
| 研究会 | CQ MIKA |
| 開催期間 | 2023-08-31 - 2023-09-01 |
| 開催地（和） | 天神岬スポーツ公園 |
| 開催地（英） | Tenjin-Misaki Sports Park |
| テーマ（和） | 無線通信品質、6G、IoT、無線伝送、リソース制御、クロスレイヤー技術、一般 |
| テーマ（英） | Wireless Communications Quality, 6G, IoT, Resource Management, Wireless Transmission, Cross layer Technologies, etc. |
| 講演論文情報の詳細 | |
| 申込み研究会 | CQ |
| 会議コード | 2023-08-CQ-MIKA |
| 本文の言語 | 英語 |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Enhancing Communication Efficiency for UAV Networks through Knowledge Distillation and Transfer Learning in Federated Learning |
| サブタイトル（英） | |
| キーワード(1)（和/英） | UAV Networks / UAV Networks |
| キーワード(2)（和/英） | Federated Learning / Federated Learning |
| キーワード(3)（和/英） | Model Switching / Model Switching |
| キーワード(4)（和/英） | Knowledge Distillation / Knowledge Distillation |
| キーワード(5)（和/英） | Transfer Learning / Transfer Learning |
| キーワード(6)（和/英） | / |
| キーワード(7)（和/英） | / |
| キーワード(8)（和/英） | / |
| 第1著者 氏名（和/英/ヨミ） | 李 亜龍 / Yalong Li / リ アリュウ |
| 第1著者 所属（和/英） | 電気通信大学 (略称： 電通大) the university of electro-communications (略称： UEC) |
| 第2著者 氏名（和/英/ヨミ） | 杜 兆陽 / Zhaoyang Du / ト チョウヨウ |
| 第2著者 所属（和/英） | 電気通信大学 (略称： 電通大) the university of electro-communications (略称： UEC) |
| 第3著者 氏名（和/英/ヨミ） | 策力 木格 / Celimuge Wu / チリ ムゲ |
| 第3著者 所属（和/英） | 電気通信大学 (略称： 電通大) the university of electro-communications (略称： UEC) |
| 第4著者 氏名（和/英/ヨミ） | 吉永 努 / Tsutomu Yoshinaga / ヨシナガ ツトム |
| 第4著者 所属（和/英） | 電気通信大学 (略称： 電通大) the university of electro-communications (略称： UEC) |
| 講演者 | 第1著者 |
| 発表日時 | 2023-08-31 16:20:00 |
| 発表時間 | 25分 |
| 申込先研究会 | CQ |
| 資料番号 | CQ2023-30 |
| 巻番号（vol） | vol.123 |
| 号番号（no） | no.174 |
| ページ範囲 | pp.26-31 |
| ページ数 | 6 |
| 発行日 | 2023-08-24 (CQ) |
[研究会発表申込システムのトップページに戻る]