Aoyu Li

Deep Learning / Computer Vision

About Me

  • Research Interests

    Computer Vision Computer Graphics

    Recently I am working on Deep Metric Learning.

Contact

  • aoyuli@rio.gsic.titech.ac.jp

Qualifications

Interests

  • Drum Set
  • Basketball

Education

  • Tokyo Institute of Technology                                               Apr. 2020-now

    YOKOTA Lab, Computer Science, Graduate School of Computing

    • Tutor: Rio Yokota; Research Area: HPC & Deep Learning

      <Non-Uniform SGD> <Deep Metric Learning> <K-FAC>

  • Sendagaya Japanese Institute                                    Oct. 2018-Mar. 2020

    Japanese language course: Advanced III

    • Scholarship for Outstanding Students                                        Jul. 2019 & Jan. 2020

  • Massachusetts Institute of Technology                                 Jul.-Sep. 2018

    Visiting student, Machine Learning Research Group, MIT CSAIL

  • Ningbo Institute of Technology, Zhejiang University   Sep. 2014-Jun. 2018

    Bachelor of Economics, School of Business; Minor: Computer Science

    • Tutor: Pei Wang; Research Area: Financial Engineering

      <Abnormality Detection> <ATE Semi-Parametric Model>

    • Outstanding Paper Award                                                                              May 2018

    • Scholarship for Outstanding Students                                      Sep. 2016 & Sep. 2017

Work Experience

  • Denso IT Lab                                                             Tokyo, Jul. 2020-now

    Recognition & Learning Algorithm Lab. (Research Assistant)

    • Implemented methods based on semantic distances for image retrieval tasks.

    • Improved performance of DML losses on unbalanced / noisy datasets by NU-SGD.

  • AIST OIL                                                                   Tokyo, Jan.-Apr. 2021

    Technical Staff, IT & Human Factors

    • Large scale DL model training and performance testing with ABCI.

Projects & Research Experience

  • Tokyo Institute of Technology, RioYokota Lab           Tokyo, Apr. 2021-now

    Second-Order Optimization (Research Assistant)

    • Research and implementation of optimization algorithms.

    • Use K-FAC-based NGD approximation algorithm to accelerate model convergence.

  • QWeather Group Project                                                     Mar.-Jun. 2021

    iOS Client

    • Mainly responsible for client side during development.

    • iOS Project: Real Time Weather

  • Behavior prediction, Unity Hub                                             Jul.-Sep. 2020

    Flocking Simulation

    • Provide experimental data for research related to behavior prediction and accurate simulation of fish flocking using Unity.

    • Unity Project: Flocking

Honors & Awards

  • Global Entrepreneur Program             New York & Boston, Jan.-Feb. 2015

    Data Analyst of group G

    • The First Runner Up, Social Entrepreneurship Competition, Harvard University

    • The First Runner Up, International Financial Planning Competition, Morgan Stanley

Skills

  • Programming Languages

    familiar with C++, Python; learn about Swift, CUDA, MPI, OpenGL

  • DevOps

    AWS, GitHub

  • Softwares

    Vim, VSCode, Xcode, Unity, Adobe AfterEffects, Final Cut Pro