I am currently a second-year Ph.D. student in Machine Learning and High-Performance Computing at Imperial College London, supervised by Prof. Wayne Luk.
My current research focuses on efficient algorithm and acceleration for Machine Learning applications. My previous research is about Quantization (Block Floating Point and Partial Quantization) and efficient structure (F-E3D) on various hardware platforms. I also have experience with machine learning on the mobile platform (Android Demo).
2018 International Conferenceon Field Programmable Logic and Applications (FPL)
Fan, Hongxiang, H.-C. Ng, and W. Luk,
2017 International Conference on Field Programmable Logic and Applications (FPL)
Fan, Hongxiang, X. Niu, Q. Liu, and W. Luk,
I was a research assistant with Prof. Qiang Liu at Tianjin University. The project (publication) is about FPGA-based hardware emulation for IC security evaluation. My responsibility was building a software tool using Python to automatically extract the necessary information from the synthesized netlist.
Under the supervision of Prof. Wayne Luk, the research studied the FPGA-based acceleration of 3D CNNs,which led to the publication in FPL’17. The internship was sponsored by Tianjin University with full scholarship.
My work focused on the optimization and acceleration of deep learning algorithm. Based on my proposed partial quantization technique, the processing speed of the face detection product increased from 30 framesper second (fps) to 100 fps with negligible accuracy loss. I also helped in designing the hardware architecture for deep learning algorithm.(publication)
Machine Learning and Natural Language Processing (NLP)