A 21mW low-power recurrent neural network accelerator with quantization tables for embedded deep learning applications | IEEE Conference Publication | IEEE Xplore

A 21mW low-power recurrent neural network accelerator with quantization tables for embedded deep learning applications


Abstract:

A 21mW low-power embedded Recurrent Neural Network (RNN) accelerator is proposed to realize the image captioning applications. The low-power RNN operation is achieved by ...Show More

Abstract:

A 21mW low-power embedded Recurrent Neural Network (RNN) accelerator is proposed to realize the image captioning applications. The low-power RNN operation is achieved by 3 key features: 1) Quantization-table-based matrix multiplication with RNN weight quantization, 2) Dynamic quantization-table allocation scheme for balanced pipelined RNN operation, and 3) Zero-skipped RNN operation using quantization-table. The Quantization table enables the 98% reduction of the multiplier operations by replacing the multiplication to the table reference. The dynamic quantization-table allocation is used to achieve high chip-utilization efficiency over 90% by balanced pipeline operation for three variations of the RNN operation. The zero-skipped RNN operation reduces the overall 27% of required external memory bandwidth and quantization-table operations without any additional hardware cost. The proposed RNN accelerator of 1.84mm2 achieves 21mW power consumption and demonstrates its functionality on the image captioning RNN in 65nm CMOS process.
Date of Conference: 06-08 November 2017
Date Added to IEEE Xplore: 28 December 2017
ISBN Information:
Conference Location: Seoul, Korea (South)

References

References is not available for this document.