Kai Xu

Kai Xu

Auto-generated profile, based on publicly available information

Kai Xu

Beihang University, Beijing, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
28 October 2024
Verified
Conference Paper
Open access
Ying LiYing Li
Ying Li
KXKai Xu
Kai Xu
YLYangdong Liu
Yangdong Liu
JLJingzhuo Liang
Jingzhuo Liang
YYYue Yan
Yue Yan
HYHao You
Hao You
With the advancement of artificial intelligence and the Internet of Things, the demand for deploying neural networks on embedded devices is steadily increasing. Field Programmable Gate Arrays (FPGAs) are an optimal solution for this challenge due to their low power consumption, low latency, and programmability, which has garnered significant attention in the industry. This paper proposes a hardware accelerator architecture that leverages both the Processing System (PS) and Programmable Logic (PL) sides in parallel, based on the ZYNQ-7000 series System on Chip (SoC), and demonstrates its superior performance through the deployment of multiple convolutional models.

28 October 2024
Verified
Conference Paper
Open access
Ying LiYing Li
Ying Li
YLYangdong Liu
Yangdong Liu
KXKai Xu
Kai Xu
JLJingzhuo Liang
Jingzhuo Liang
YYYue Yan
Yue Yan
HYHao You
Hao You
FPGA devices, known for their low power consumption, are ideal for large-scale computations. Neural networks, which demand substantial computational power, can benefit from FPGA acceleration, particularly in embedded systems. This paper investigates FPGA's potential to enhance neural network performance on embedded devices through circuit design and rapid hardware deployment. Unlike PCs, embedded systems need specialized tools for efficient research and deployment. The study examines operator design, utilizing techniques like multiply-accumulator (MAC) and systolic arrays, to develop an automatic generation tool for reusable operators. These operators adapt to various input sizes, maximizing FPGA resource efficiency. Experiments confirm the operators' correctness, deployability, and acceleration, demonstrating superior performance compared to CPUs of the same frequency, thereby effectively supporting neural network deployment in embedded fields.