In order for deep learning applications to run efficiently on low-power edge devices, including mobile and internet-of-things systems, it is important to reduce their computational and memory requirements. Binarized neural networks have shown promise in this area, but these are typically designed using existing architectures based on floating-point number representations. A more promising approach is to apply network architecture search algorithms to find optimized binarized architectures. In this paper, encoding schemes for the genetic algorithm search of binarized networks are described. The simulation results demonstrate the effectiveness of the proposed method.
|Original language||English (US)|
|Title of host publication||Proceedings - 2019 IEEE 13th International Conference on ASIC, ASICON 2019|
|Editors||Fan Ye, Ting-Ao Tang|
|Publisher||IEEE Computer Society|
|State||Published - Oct 2019|
|Event||13th IEEE International Conference on ASIC, ASICON 2019 - Chongqing, China|
Duration: Oct 29 2019 → Nov 1 2019
|Name||Proceedings of International Conference on ASIC|
|Conference||13th IEEE International Conference on ASIC, ASICON 2019|
|Period||10/29/19 → 11/1/19|
Bibliographical noteFunding Information:
This work was supported in part by Fudan University State Key Lab of ASIC and Systems, under Grant 2018GF016.