A Low Power DNN-based Speech Recognition Processor with Precision Recoverable Approximate Computing

Created by W.Langdon from gp-bibliography.bib Revision:1.7810

@InProceedings{Liu:2022:ISCAS,

• author = "Bo Liu2 and Xuetao Wang and Renyuan Zhang and Anfeng Xue and Ziyu Wang and Haige Wu and Hao Cai",
• booktitle = "2022 IEEE International Symposium on Circuits and Systems (ISCAS)",
• title = "A Low Power {DNN-based} Speech Recognition Processor with Precision Recoverable Approximate Computing",
• year = "2022",
• pages = "2102--2106",
• abstract = "This paper proposes a low power speech recognition processor based on an optimized DNN with precision recoverable approximate computing. In order to accelerate and improve energy of DNN, an approximate multiplier based on cartesian genetic programming with weight pre-classification and mismatch compensation is proposed. A partial retraining scheme based on approximate noise is proposed to recover the accuracy loss caused by approximate computing. Experimental results show that the proposed approximate multiplier reduces power consumption by 42.percent, and the partial retraining scheme can recover accuracy of 3.0percent~4.3percent. Implemented under 22nm, the proposed processor can support the recognition of 10 keywords under different noise types and signal-to-noise ratios (5dB~clean), while the recognition accuracy is 83.3percent ~89.8percent and power consumption is 8.6μ W.",
• keywords = "genetic algorithms, genetic programming, Cartesian genetic programming",
• DOI = "doi:10.1109/ISCAS48785.2022.9937896",
• ISSN = "2158-1525",
• month = may,
• notes = "Also known as \cite{9937896}",

}

Genetic Programming entries for Bo Liu2 XueTao Wang RenYuan Zhang AnFeng Xue Ziyu Wang Haige Wu Hao Cai

Citations