This paper proposes a hardware-efficient architecture, Linearized Convolution Network (LiCo-Net) for keyword spotting. It is optimized specifically for low-power processor units like microcontrollers. ML operators exhibit heterogeneous efficiency profiles on power-efficient hardware. Given the exact theoretical computation cost, int8 operators are more computation-effective than float operators, and linear layers are often more efficient than other layers. The proposed LiCo-Net is a dual-phase system that uses the efficient int8 linear operators at the inference phase and applies streaming convolutions at the training phase to maintain a high model capacity. The experimental results show that LiCo-Net outperforms single-value decomposition filter (SVDF) on hardware efficiency with on-par detection performance. Compared to SVDF, LiCo-Net reduces cycles by 40% on HiFi4 DSP.
@article{arxiv.2211.04635,
title = {LiCo-Net: Linearized Convolution Network for Hardware-efficient Keyword Spotting},
author = {Haichuan Yang and Zhaojun Yang and Li Wan and Biqiao Zhang and Yangyang Shi and Yiteng Huang and Ivaylo Enchev and Limin Tang and Raziel Alvarez and Ming Sun and Xin Lei and Raghuraman Krishnamoorthi and Vikas Chandra},
journal= {arXiv preprint arXiv:2211.04635},
year = {2022}
}