Name
Title | ||
|---|---|---|
An energy-efficient voice activity detector using deep neural networks and approximate computing |
Abstract | ||
|---|---|---|
This paper proposed an energy-efficient reconfigurable DNN accelerator architecture for voice activity detection (VAD) based on deep neural networks and fabricated in 28-nm technology. To reduce the power consumption and achieve high energy efficiency, two optimization techniques are proposed. First, the processing elements contained in the DNN accelerator support digital-analog mixed approximate computing, including multi-step quantized multiplication units and time-delay based addition units. Second, the proposed approximate computing units can be dynamically reconfigured to adapt to different computing accuracy requirements. The proposed approximate computing can significantly reduce the power consumption by 76% ∼ 88% compared to standard digital computing units. Implemented under TSMC 28 nm HPC + process technology, the layout size of the prototype system is 0.52 mm2, and the estimated power is 6 ∼ 12 μW. The energy efficiency of our work achieves 33.33 ∼ 66.67 TOPS/W, which is over 6.5X better than the state-of-the-art architecture. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1016/j.mejo.2019.03.009 | Microelectronics Journal |
Keywords | Field | DocType |
Voice activity detection,Deep neural networks,Approximate computing | Efficient energy use,Voice activity detection,Electronic engineering,Multiplication,Quantization (physics),Engineering,Detector,Deep neural networks,TOPS,Approximate computing | Journal |
Volume | ISSN | Citations |
87 | 0026-2692 | 0 |
PageRank | References | Authors |
0.34 | 0 | 7 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Bo Liu | 1 | 6 | 5.82 |
| Zhen Wang | 2 | 76 | 28.98 |
| Shisheng Guo | 3 | 0 | 0.68 |
| Huazhen Yu | 4 | 0 | 0.34 |
| Yu Gong | 5 | 12 | 7.36 |
| Jun Yang | 6 | 588 | 39.42 |
| Longxing Shi | 7 | 116 | 39.08 |