Bin Zhao

Bin Zhao

Auto-generated profile, based on publicly available information

Bin Zhao

Department of IC Design, Institute of Microelectronics, A\ast STAR, Singapore 138643

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 May 2019 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
KNKian Ann Ng
Kian Ann Ng
CYChao Yuan
Chao Yuan
ARAstrid Rusly
Astrid Rusly
ADAnh-Tuan Do
Anh-Tuan Do
BZBin Zhao
Bin Zhao
YXYong Ping Xu
Yong Ping Xu
The adoption of chronic implantable peripheral nerve-based prosthetic devices is currently hampered by the lack of a highly integrated neural signal acquisition system-on-chip (SoC). We report a ten-channel peripheral nervous system (PNS) electroneurogram (ENG) signal acquisition SoC within an implantable package. Requiring only four off-chip capacitors, this SoC can be co-encapsulated with flexible nerve electrodes and a resonant coil antenna to form a 3.4 cm³ and 3.9 g implantable device for chronic ENG acquisition. This SoC is inductively powered and controlled through a resonant coil at 22 MHz and transmits the digitized neural signal through a near-infrared LED (NIR-LED). Fabricated in 0.18-μm CMOS, each amplifier channel exhibits an input referred noise of 1.9 μVrms and a noise efficiency factor (NEF) of 4.0 within the signal bandwidth of 5.5 kHz. Each amplifier channel within the SoC is digitized with 10-bit resolution at 17.5 ksps, and the total power consumption (SoC and NIR-LED) is 4.4 mW when the NIR-LED is driven at 3 Mb/s. An electrode impedance measurement circuit with <10% magnitude and <8° angle error for measuring impedances up to 1 MΩ is also incorporated in this SoC. This wireless, low noise ENG acquisition SoC package has been validated in vivo while implanted on a rodent to acquire ENG from its sciatic nerve.

4 January 2019 in arXiv
Verified
Preprint
Open access
PDF file
XGXue Geng
Xue Geng
JFJie Fu
Jie Fu
BZBin Zhao
Bin Zhao
JLJie Lin
Jie Lin
MAMohamed M. Sabry Aly
Mohamed M. Sabry Aly
VCVijay Chandrasekhar
Vijay Chandrasekhar
This paper addresses a challenging problem - how to reduce energy consumption without incurring performance drop when deploying deep neural networks (DNNs) at the inference stage. In order to alleviate the computation and storage burdens, we propose a novel dataflow-based joint quantization approach with the hypothesis that a fewer number of quantization operations would incur less information loss and thus improve the final performance. It first introduces a quantization scheme with efficient bit-shifting and rounding operations to represent network parameters and activations in low precision. Then it restructures the network architectures to form unified modules for optimization on the quantized model. Extensive experiments on ImageNet and KITTI validate the effectiveness of our model, demonstrating that state-of-the-art results for various tasks can be achieved by this quantized model. Besides, we designed and synthesized an RTL model to measure the hardware costs among various quantization methods. For each quantization operation, it reduces area cost by about 15 times and energy consumption by about 9 times, compared to a strong baseline.

5 October 2020 in International Journal of Machine Learning and Computing
Verified
Journal Article
Open access
KAKong Anmin
Kong Anmin
ZBZhao Bin
Zhao Bin