Optimizing Edge AI: A Comprehensive Survey on Data, Model, and System Strategies * Corresponding author
Xubin Wang 1,2,3 , Weijia Jia 2,3∗ · 2025
Read Optimizing Edge AI: A Comprehensive Survey on Data, Model, and System Strategies * Corresponding author by Xubin Wang 1,2,3 , Weijia Jia 2,3∗ (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-03265.txt.
Description
Xubin Wang, Qing Li, and Weijia Jia survey how to run large AI models, in a 2025 preprint, on small devices at the edge of a network. Big language models and AI agents need lots of memory and compute. Phones and other edge devices have little of either. The survey lays out a framework for keeping the models' reasoning power while shrinking the cost.
The authors group methods into three layers. The first shrinks the model itself through tricks like quantization, pruning, and distillation. The second covers system design: running on the device, shifting work to the cloud, or mixing both. The third adapts the work to each task and user. They also propose a shared way to score systems on speed, energy, accuracy, and privacy. Open problems include fair tests and clear energy reporting. The paper ends with design advice for builders.
Catalog record
| CREATOR | Xubin Wang 1,2,3 , Weijia Jia 2,3∗ |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 133 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-03265