Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2404.09544 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910410224959488 |
|---|---|
| author | Qiao, Tong Yang, Jianlei Qi, Yingjie Zhou, Ao Bai, Chen Yu, Bei Zhao, Weisheng Hu, Chunming |
| author_facet | Qiao, Tong Yang, Jianlei Qi, Yingjie Zhou, Ao Bai, Chen Yu, Bei Zhao, Weisheng Hu, Chunming |
| contents | Graph Neural Networks (GNNs) succeed significantly in many applications recently. However, balancing GNNs training runtime cost, memory consumption, and attainable accuracy for various applications is non-trivial. Previous training methodologies suffer from inferior adaptability and lack a unified training optimization solution. To address the problem, this work proposes GNNavigator, an adaptive GNN training configuration optimization framework. GNNavigator meets diverse GNN application requirements due to our unified software-hardware co-abstraction, proposed GNNs training performance model, and practical design space exploration solution. Experimental results show that GNNavigator can achieve up to 3.1x speedup and 44.9% peak memory reduction with comparable accuracy to state-of-the-art approaches. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_09544 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | GNNavigator: Towards Adaptive Training of Graph Neural Networks via Automatic Guideline Exploration Qiao, Tong Yang, Jianlei Qi, Yingjie Zhou, Ao Bai, Chen Yu, Bei Zhao, Weisheng Hu, Chunming Machine Learning Artificial Intelligence Graph Neural Networks (GNNs) succeed significantly in many applications recently. However, balancing GNNs training runtime cost, memory consumption, and attainable accuracy for various applications is non-trivial. Previous training methodologies suffer from inferior adaptability and lack a unified training optimization solution. To address the problem, this work proposes GNNavigator, an adaptive GNN training configuration optimization framework. GNNavigator meets diverse GNN application requirements due to our unified software-hardware co-abstraction, proposed GNNs training performance model, and practical design space exploration solution. Experimental results show that GNNavigator can achieve up to 3.1x speedup and 44.9% peak memory reduction with comparable accuracy to state-of-the-art approaches. |
| title | GNNavigator: Towards Adaptive Training of Graph Neural Networks via Automatic Guideline Exploration |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2404.09544 |