Zhengyu Yang

Zhengyu Yang

Research Assistant

University of Southern California

About Me

I am a full-time research assistant at the University of Southern California Viterbi School of Engineering.

Interests
  • Computer Vision
  • Federated Learning
  • Self-Supervised Learning
  • Optimization
  • Statistical Learning
  • Reinforcement Learning
Education
  • MS in Computer Science, 2021

    University of Southern California

  • BS in Computer Science, 2021

    University of Southern California

  • BS in Applied Mathematics, 2021

    University of Southern California

Experience

 
 
 
 
 
Research Assistant
Dec 2020 – Present California
  • Advised by Professor Salman A. Avestimehr
  • Investigated areas:
    • Federated Learning
    • Distributed Optimization
  • Projects:
    • Self-Supervised Federated Learning With Personalization Project (under review for NeurIPS 2021)
 
 
 
 
 
Research Assistant
Aug 2019 – Present California
  • Advised by Professor Ramakant Nevatia
  • Investigated areas:
    • Computer Vision
    • Video
    • Semi-Supervised/Self-Supervised Learning
    • Low Shot Learning
  • Projects:
    • SimPLE: Similar Pseudo Label Exploitation for Semi-Supervised Classification (co-first authored, published for CVPR 2021)
    • DARPA Learning with Less Label Project: Image Classification and Video Action Recognition
    • Self-Supervised Learning Project (to submit for ICLR 2022)
 
 
 
 
 
Undergraduate Research Assistant
Feb 2019 – Aug 2019 California
  • Advised by Professor Joseph Lim
  • Investigated areas:
    • Robot Learning
    • Reinforcement Learning
    • Learning From Demonstration
  • Projects:
    • IKEA Furniture Assembly Environment for Long-Horizon Complex Manipulation Tasks
    • To Follow or not to Follow: Selective Imitation Learning from Observation
 
 
 
 
 
Software Engineer Intern
May 2016 – Jul 2016 Shanghai
Developed a server program to collect pedometer statistics and then ranked and displayed them in internal social accounts.

Contact