Tram-FL: Routing-based Model Training for Decentralized Federated Learning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914990696431616 |
|---|---|
| author | Maejima, Kota Nishio, Takayuki Yamazaki, Asato Hara-Azumi, Yuko |
| author_facet | Maejima, Kota Nishio, Takayuki Yamazaki, Asato Hara-Azumi, Yuko |
| contents | In decentralized federated learning (DFL), substantial traffic from frequent inter-node communication and non-independent and identically distributed (non-IID) data challenges high-accuracy model acquisition. We propose Tram-FL, a novel DFL method, which progressively refines a global model by transferring it sequentially amongst nodes, rather than by exchanging and aggregating local models. We also introduce a dynamic model routing algorithm for optimal route selection, aimed at enhancing model precision with minimal forwarding. Our experiments using MNIST, CIFAR-10, and IMDb datasets demonstrate that Tram-FL with the proposed routing delivers high model accuracy under non-IID conditions, outperforming baselines while reducing communication costs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2308_04762 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Tram-FL: Routing-based Model Training for Decentralized Federated Learning Maejima, Kota Nishio, Takayuki Yamazaki, Asato Hara-Azumi, Yuko Machine Learning Distributed, Parallel, and Cluster Computing Networking and Internet Architecture 68T20 (Primary) 68M14 (Secondary) I.2.7; I.2.8; I.2.10; I.2.11 In decentralized federated learning (DFL), substantial traffic from frequent inter-node communication and non-independent and identically distributed (non-IID) data challenges high-accuracy model acquisition. We propose Tram-FL, a novel DFL method, which progressively refines a global model by transferring it sequentially amongst nodes, rather than by exchanging and aggregating local models. We also introduce a dynamic model routing algorithm for optimal route selection, aimed at enhancing model precision with minimal forwarding. Our experiments using MNIST, CIFAR-10, and IMDb datasets demonstrate that Tram-FL with the proposed routing delivers high model accuracy under non-IID conditions, outperforming baselines while reducing communication costs. |
| title | Tram-FL: Routing-based Model Training for Decentralized Federated Learning |
| topic | Machine Learning Distributed, Parallel, and Cluster Computing Networking and Internet Architecture 68T20 (Primary) 68M14 (Secondary) I.2.7; I.2.8; I.2.10; I.2.11 |
| url | https://arxiv.org/abs/2308.04762 |