DANCE: Resource-Efficient Neural Architecture Search with Data-Aware and Continuous Adaptation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911126525050880 |
|---|---|
| author | Wang, Maolin Wei, Tianshuo Zhang, Sheng Guo, Ruocheng Wang, Wanyu Ye, Shanshan Zou, Lixin Wei, Xuetao Zhao, Xiangyu |
| author_facet | Wang, Maolin Wei, Tianshuo Zhang, Sheng Guo, Ruocheng Wang, Wanyu Ye, Shanshan Zou, Lixin Wei, Xuetao Zhao, Xiangyu |
| contents | Neural Architecture Search (NAS) has emerged as a powerful approach for automating neural network design. However, existing NAS methods face critical limitations in real-world deployments: architectures lack adaptability across scenarios, each deployment context requires costly separate searches, and performance consistency across diverse platforms remains challenging. We propose DANCE (Dynamic Architectures with Neural Continuous Evolution), which reformulates architecture search as a continuous evolution problem through learning distributions over architectural components. DANCE introduces three key innovations: a continuous architecture distribution enabling smooth adaptation, a unified architecture space with learned selection gates for efficient sampling, and a multi-stage training strategy for effective deployment optimization. Extensive experiments across five datasets demonstrate DANCE's effectiveness. Our method consistently outperforms state-of-the-art NAS approaches in terms of accuracy while significantly reducing search costs. Under varying computational constraints, DANCE maintains robust performance while smoothly adapting architectures to different hardware requirements. The code and appendix can be found at https://github.com/Applied-Machine-Learning-Lab/DANCE. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_04671 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | DANCE: Resource-Efficient Neural Architecture Search with Data-Aware and Continuous Adaptation Wang, Maolin Wei, Tianshuo Zhang, Sheng Guo, Ruocheng Wang, Wanyu Ye, Shanshan Zou, Lixin Wei, Xuetao Zhao, Xiangyu Machine Learning Computer Vision and Pattern Recognition Neural Architecture Search (NAS) has emerged as a powerful approach for automating neural network design. However, existing NAS methods face critical limitations in real-world deployments: architectures lack adaptability across scenarios, each deployment context requires costly separate searches, and performance consistency across diverse platforms remains challenging. We propose DANCE (Dynamic Architectures with Neural Continuous Evolution), which reformulates architecture search as a continuous evolution problem through learning distributions over architectural components. DANCE introduces three key innovations: a continuous architecture distribution enabling smooth adaptation, a unified architecture space with learned selection gates for efficient sampling, and a multi-stage training strategy for effective deployment optimization. Extensive experiments across five datasets demonstrate DANCE's effectiveness. Our method consistently outperforms state-of-the-art NAS approaches in terms of accuracy while significantly reducing search costs. Under varying computational constraints, DANCE maintains robust performance while smoothly adapting architectures to different hardware requirements. The code and appendix can be found at https://github.com/Applied-Machine-Learning-Lab/DANCE. |
| title | DANCE: Resource-Efficient Neural Architecture Search with Data-Aware and Continuous Adaptation |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.04671 |