Master student in Computer Science, University of Southern California
June, 2017 - Present
B. S. in Computer Science and Technology, Peking University
Sep, 2013 - July, 2017
Summer Research Assistant, Zhang Research Group, Computer Systems Lab, Cornell University
July, 2016 - Sept, 2016 2 months
Did some research on Binarized Neural Network (BNN) Acceleration.
Against x86 processor, Intel Xeon E5-2640, our group achieved a 2.5x speedup with FPGA accelerator
on Zedboard, a platform with much lower power consumption.
I rebuilt the BNN in C++ (link), including dense-layer, conv-layer, pool-layer and batch-norm-layer, and then parallelized BNN pipeline with Halide, tested it on x86 processor and mGPU board, got a 2x speedup on average against the BNN implementation with Lasagne.
It was a really wonderful experience to work with PhD student Ritchie Zhao, Steve Dai and Professor Zhiru Zhang. After this experience, I was proficient in working on a sever, and developing in Linux environment. Also, I learned how to collaborate with partners in a team project.
Software Development Intern, National Engineering Lab for Video Technology, Peking University
Apr, 2015 - June, 2016 1 year 2 months
My ex-roommate and good friend, Yuhang Zhao and I developed a video searching demo based on image feature during the research internship.
We extracted the image feature with SIFT algorithm, an OpenCV built-in function. I introduced Compact Descriptors for Visual Search (CDVS) as an efficient method extracting the video frame features, and designed an image-to-video matching scheme. Besides the basic design, we experimented different parameter values for frame downsampling and video interval sampling to gain speedup without loss of accuracy.
Frankly speaking, the searching speed of the demo was reasonably slow, though we had used the cutting-edge technique CDVS provided by Ph.D. student Jie Chen, though I dived myself into the searching algorithm and improved it as much as possible. Because of that, I knew that research was full of difficulties, so I really respect the researchers who devote themselves in science and engineering.
By the way, thanks for the guidance from Professor Ma. He inspired me a lot during that year, and he was willing to answer my questions and help me in both study and life.
Stock Search Web App
Nov, 2017
Developed a website and android application that allows users to search stock information using the Alpha Vantage API, manage and share their favorite stock.
Web front end: Check it out at this link
and Github link
Android front end: Download
and Github link
Back end: Node.js
Route Recognition Based on Coarse-grained Cellphone Location Information
Apr, 2016 - May, 2017
Applied a multi-layer dynamic programming algorithm to retrieve the route based on simulative location data. (report in Chinese)
Log File System
Jan, 2016
Wrapped file modifying APIs in HDFS and constructed a log file system. (report in Chinese)
Pedestrian and Car Detection on PKUSVD Dataset
Nov, 2015 - Dec, 2015
Applied Fast-RCNN model on PKUSVD Dataset, got 85% precision with 80% recall on average. (report)
Wiki News Classification based on Bayes
Nov, 2015
Crawled web data from Wiki News with Python framework, Scrapy, and classified the news into several categories. (report in Chinese)
Image Segmentation
June, 2015
Implemented semi-supervised image segmentation based on min-cut algorithm. (report)
Gold Medal (Tenth Place), The ACM-ICPC Asia Regional Contest Mudanjiang Site 2014
Silver Medal (Twenty-third Place), The ACM-ICPC Asia Regional Contest Chengdu Site 2013
International Olympiad in Informatics (IOI) 2013 China Team Selection Competition Ninth Place
Merit Student of Peking University, 2013 - 2014 Acadamic Year
Tung OOCL Scholarship, Peking University, 2013 - 2014 Acadamic Year
May Fourth Scholarship, Peking University, 2015 - 2016 Acadamic Year
Learning Improvement Award, Peking University, 2015 - 2016 Acadamic Year
Zhao, Ritchie, Weinan Song, Wentao Zhang, Tianwei Xing, Jeng-Hau Lin, Mani B. Srivastava, Rajesh Gupta, and Zhiru Zhang. ”Accelerating Binarized Convolutional Neural Networks with SoftwareProgrammable FPGAs.” In FPGA, pp. 15-24. 2017