Separation of Lanthanides/ Lanthanides and Actinides
Yixin Jin Resume - Computer Science and Mathematics Student
1. Yixin Jin
Address:
1760 Broadway Street, Apt 338
Ann Arbor, MI 48105
Email:jinyixin@umich.edu
Mobile: 734-353-9769
Website: www-personal.umich.edu/~jinyixin
EDUCATION University of Michigan, Ann Arbor, MI
B.S. in Computer Science and Mathematics, 2013-2016 GPA: 3.93
Zhejiang University of Finance & Economics, Hangzhou, China
B.Ec. in International Economics & Trade, Minor in Finance, 2010-2013 GPA: 3.93
RESEARCH
EXPERIENCE
Summer Undergraduate Research in
Engineering
Supervised by Prof.Jia Deng
UM Vision&Learning Lab, Ann Arbor, MI May 2015 - Present
Research on human pose estimation using convolutional neural network.
Implemented a modified version of inception module for our designed network.
Adopted Batch Normalization method which I read from a paper to accelerate training.
Explored preprocessing methods for input data and new architectures of network to
improve the performance of our system.
WORK
EXPERIENCE
Grader Assisted Prof.Shifi Reif
University of Michigan, Ann Arbor, MI Jan 2015 - May 2015
Gave solutions and graded weekly homework of undergraduate level linear algebra
HONORS &
AWARDS
University Honors 2013-2015
James B. Angell Scholar 2015
EECS Scholar 2015
Margaret S. Huntington Scholarship 2013
ZUFE First-Class Academic Excellence Scholarship(top 3% students) 2012
SELECTED
PROJECTS
Accelerating Deep Network Training Using Batch Normalization:
• Implemented Batch Normalization Method as described in the paper using CAFFE
• Trained some classical deep networks such as GoogLeNet, AlexNet showing the ef-
fectiveness of this method.
Right-Whale Recognition:
• Inspired by the paper DeepFace:Closing the Gap to Human-Level Performance in
Face Verification.
• Designed a system which first accurately detected the Right Whale and cropped the
Right Whale in the image using a fine-tuned AlexNet.Took the cropped image as
input, I trained a deep neural network classifier to recognize different Right Whales.
Error Analysis on Object Scene Flow for Autonomous Vehicles:
• Replicated the experiments in the paper Object Scene Flow for Autonomous Vehicles.
• Found out when the input images with horrible lighting condition or containing
objects which had large movements, the system failed to generate the correct object
map.
• In order to generate the correct object map, we fixed the object detection step by
identifying the objects with large movement and showed quantitative improvements
in terms of their evaluation metrics.
COMPUTER
SKILLS
Languages: C, C++, Python, MATLAB, Bash, Assembly, SQL, LATEX.
Tools & Libary: OpenCV, CAFFE, Torch7