Title | ||
---|---|---|
EERA-ASR: An Energy-Efficient Reconfigurable Architecture for Automatic Speech Recognition With Hybrid DNN and Approximate Computing. |
Abstract | ||
---|---|---|
This paper proposes a hybrid deep neural network (DNN) for automatic speech recognition and an energy-efficient reconfigurable architecture with approximate computing for accelerating the DNN. To accelerate the hybrid DNN and reduce the energy consumption, we propose a digital-analog mixed reconfigurable architecture with approximate computing units, including a binary weight network accelerator with analog multi-chain delay-addition units for bit-wise approximate computing and a recurrent neural network accelerator with approximate multiplication units for different calculation accuracy requirements. Implemented under TSMC 28nm HPC+ process technology, the proposed architecture can achieve the energy efficiency of 163.8TOPS/W for 20 keywords recognition and 3.3TOPS/W for common speech recognition. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/ACCESS.2018.2870273 | IEEE ACCESS |
Keywords | Field | DocType |
Hybrid deep neural network,binary weight network,reconfigurable architecture,approximate computing | Adder,Computer science,Efficient energy use,Field-programmable gate array,Recurrent neural network,Speech recognition,Multiplication,Artificial neural network,Energy consumption,Binary number | Journal |
Volume | ISSN | Citations |
6 | 2169-3536 | 0 |
PageRank | References | Authors |
0.34 | 0 | 6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Bo Liu | 1 | 10 | 4.67 |
Hai Qin | 2 | 0 | 1.01 |
Yu Gong | 3 | 12 | 7.36 |
Wei Ge | 4 | 21 | 11.72 |
Mengwen Xia | 5 | 0 | 1.01 |
Longxing Shi | 6 | 116 | 39.08 |