LIYUAN LIU

Prof. LIYUAN LIU

Institute of Semiconductors, Chinese Academy of Science, Beijing, China

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

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

Representative Publications

View All Publications
1 January 2021
Verified
Conference Paper
Open access
PDF file
XGXiaotao Gu
Xiaotao Gu
LLLIYUAN LIU
LIYUAN LIU
HYHongkun Yu
Hongkun Yu
JLJing Li
Jing Li
CCChen Chen
Chen Chen
JHJiawei Han
Jiawei Han
Xiaotao Gu, Liyuan Liu, Hongkun Yu, Jing Li, Chen Chen, Jiawei Han. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

8 July 2022 in IEEE Transactions on Biomedical Circuits and Systems
Verified
Journal Article
Closed access
HWHaibing Wang
Haibing Wang
ZHZhen He
Zhen He
TWTengxiao Wang
Tengxiao Wang
JHJunxian He
Junxian He
XZXichuan Zhou
Xichuan Zhou
CSCong Shi
Cong Shi
Human brain cortex acts as a rich inspiration source for constructing efficient artificial cognitive systems. In this paper, we investigate to incorporate multiple brain-inspired computing paradigms for compact, fast and high-accuracy neuromorphic hardware implementation. We propose the TripleBrain hardware core that tightly combines three common brain-inspired factors: the spike-based processing and plasticity, the self-organizing map (SOM) mechanism and the reinforcement learning scheme, to improve object recognition accuracy and processing throughput, while keeping low resource costs. The proposed hardware core is fully event-driven to mitigate unnecessary operations, and enables various on-chip learning rules (including the proposed SOM-STDP & R-STDP rule and the R-SOM-STDP rule regarded as the two variants of our TripleBrain learning rule) with different accuracy-latency tradeoffs to satisfy user requirements. An FPGA prototype of the neuromorphic core was implemented and elaborately tested. It realized high-speed learning (1349 frame/s) and inference (2698 frame/s), and obtained comparably high recognition accuracies of 95.10%, 80.89%, 100%, 94.94%, 82.32%, 100% and 97.93% on the MNIST, ETH-80, ORL-10, Yale-10, N-MNIST, Poker-DVS and Posture-DVS datasets, respectively, while only consuming 4146 (7.59%) slices, 32 (3.56%) DSPs and 131 (24.04%) Block RAMs on a Xilinx Zynq-7045 FPGA chip. Our neuromorphic core is very attractive for real-time resource-limited edge intelligent systems.

28 December 2021 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
QLQiwen Liao
Qiwen Liao
YZYuguang Zhang
Yuguang Zhang
SMSiyuan Ma
Siyuan Ma
LWLei Wang
Lei Wang
LLLeliang Li
Leliang Li
NQNan Qi
Nan Qi
This article presents a 50-Gb/s optical transmitter (TX), consisting of a 40-nm distributed CMOS driver and a 180-nm silicon-photonic modulator. A lumped-segment Mach–Zehnder modulator (LS-MZM) is developed for high bandwidth (BW) four-level pulse amplitude (PAM-4) modulation. A multi-segment driver with limiting outputs is co-designed, which is distributed into each LS-MZM segment. By grouping these LS-MSM segments in a thermometer code, high-linearity modulation is realized without the need of power-hungry high-swing linear drivers. To improve the optical PAM-4 signal integrity, in-segment multiplexing along with clock phase interpolation is adopted to synchronize the electrical and optical signals across all segments. The hybrid coupling between the driver and modulator is devised to boost the BW of the high-speed data path, while a half-rate clock and data recovery (CDR) circuit is integrated to remove the accumulated jitter. Measurements show that the TX exhibits an extinction ratio (ER) of up to 9.8 dB and a 0.99 ratio of the level mismatch. A figure-of-merit (FoM) of 1.39 pJ/bit/dB corresponds to a 682-mW power, which can be further reduced by 40%, at the cost of a degraded ER of 4 dB. The PAM-4 CDR helps to achieve $< 10^{-12}$ BER and >0.1- $\text{U}I_{\mathrm {pp}}$ jitter tolerance (JTOL) from 10 to 100 MHz.

20 December 2023 in ACS Applied Materials & Interfaces
Verified
Journal Article
Closed access
JYJie Yang
Jie Yang
LLLIYUAN LIU
LIYUAN LIU
Di ZhangDi Zhang
Di Zhang
ZHzhang hongli
zhang hongli
JMJianhua Ma
Jianhua Ma
CWChen Wang
Chen Wang

8 October 2019 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
SMSong Ma
Song Ma
LLLIYUAN LIU
LIYUAN LIU
TFTong Fang
Tong Fang
JLJian Liu
Jian Liu
NWNANJIAN WU
NANJIAN WU
This article presents a discrete-time second-order ΔΣ modulator for the audio applications. In this modulator, a novel dynamic amplifier is proposed to realize the switched-capacitor (SC) integrators. To eliminate the common-mode (CM) voltage drop in a closed-loop dynamic amplifier during the integration phase, without the use of additional load capacitance, the reset method for the amplifier is modified. Two auxiliary branches are introduced to enhance the settling speed of the integrator. Two different flicker noise reduction techniques (FNRTs) are developed to improve the signal-to-noise-and-distortion ratio (SNDR) (about 2 dB in the audio bandwidth). The prototype modulator is fabricated in 65-nm CMOS technology with a 0.12-mm² core area, which achieves a dynamic range (DR) of 91 dB and a peak SNDR of 89.6 dB in the 24-kHz signal bandwidth. It consumes only 49-μW power from a 0.8-V supply, translating into a Schreier figure of merit (FoM) of 176.5 dB.

9 June 2022 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
JHJian He
Jian He
DLDonglai Lu
Donglai Lu
Haiyun XueHaiyun Xue
Haiyun Xue
SCSikai Chen
Sikai Chen
HLHan Liu
Han Liu
NQNan Qi
Nan Qi
This paper presents a hybrid-integrated optical transceiver front-end for beyond-400G short-reach optical links. A pair of the monolithic 8-channel laser drivers and the trans-impedance amplifier (TIA) is developed in 180nm SiGe BiCMOS, incorporating arrayed Vertical-Cavity-Surface- Emitting Lasers and photo-detectors. The driver uses a $2^{\mathrm {nd}}$ -order continuous-time linear equalizer (CTLE) to compensate for the channel loss with a nonlinear frequency response. Both the inductive peaking and RC-degeneration are embedded at the output stage to extend the optical modulation bandwidth (BW). The series-peaking and multi-stage distributed CTLE are combined in a resistive feedback TIA topology for improved BW and linearity. Measurement results show up to 100-Gb/s PAM-4 electrical eyes of the driver and TIA. The optical transmitter front-end operates 56 Gb/s, 4.1-dB extinction ratio, and 6.6-pJ/bit power efficiency, while the optical receiver front-end achieves 56-Gb/s, $10^{-6}$ bit error rate, and 5.9-pJ/bit power efficiency.