About Me

I am Zifeng Wang, a third-year PhD student in Machine Learning at Northeastern University, advised by Prof. Jennifer G. Dy. I got my BS degree in Electronic Engineering from Tsinghua University, Beijing. My study and research interests mainly focus on machine learning, computer vision and various application of artificial intelligence. I believe AI would change the world ultimately.

Before entering Northeastern, I worked with Prof. Jiwen Lu on multi-object tracking by reinforcement learning. At summer 2017, I went to University of Michigan as a visiting researcher under the instruction of Prof. Jia Deng. Before that, I worked with Prof Shengjin Wang and Professor Yong Li on vision and big data respectively. During my exchange at Northeastern University, I worked with Prof. Jennifer G. Dy on machine learning. You could refer my curriculum vitae for further information.

Education Background

  • Sep 2018 - Present, Northeastern University,
    PhD Student in Machine Learning, Dean's Fellow
  • Aug 2014 - July 2018, Tsinghua University,
    B.S. degree in Electronic Engineering, GPA 92/100 (Top 5%)
  • July 2017 - Sep 2017, University of Michigan,
    Visiting researcher at Department of Computer Science and Engineering
  • Sep 2016 - Dec 2016, Northeastern University,
    Exchange student at Department of Computer and Electrical Engineering

Selected Projects

I have joined several research teams, from big data mining, computer vision to machine learning. These precious experiences taught me a lot.

Learn-Prune-Share for Lifelong Learning

Northeastern University, Advisor: Jennifer Dy
  • Incoporated the state-of-the-art Alternating Direction Method of Multipliers (ADMM) based pruning strategy to solve the lifelong learning problem.
  • Designed a novel knowledge sharing scheme, which learns to select useful knowledge from old tasks and adapt them to the current task.
  • Paper accepted at ICDM 2020, Learn-Prune-Share for Lifelong Learning
  • .

Class Discovery Kernel Network

Northeastern University, Advisor: Jennifer Dy
  • Proposed a deep learning framework which utilizes HSIC to bridge supervised and unsupervised information together in a systematic way.
  • Addresses the problem of overfitting to old classes, leading to improved class discovery, and can be generically applied to a broad array of deep architectures.
  • Best Paper Candidate at ICDM 2020, Open-World Class Discovery with Kernel Networks
  • .

Multi-object Tracking

Tsinghua University, Advisor: Jiwen Lu

Selected Publications

Conference Papers

Journal Papers

Awards

  • Best Paper Candidate, ICDM 2020
  • Best Paper Award, DySPAN 2019
  • Travel Award, DySPAN 2019
  • Travel Award, NeurIPS 2019
  • Dean's Fellowship, 2018
    Highest honor awarded to new PhD students in Northeastern University for their out standing academic background
  • Evergrande Scholarship, 2016
    Awarded to students with excellent academic performance, scientific potential and overall development. For about top 5% out of 233 students
  • First Prize in Beijing Undergraduate Physics Competition, 2015
    High-level Physics competition, awarded to top 3% students
  • Ranked 3th at 17th Electronic Design Contest, 2015
    Top robotic contest at Tsinghua University. Top 1% out of about 400 teams

Skills

  • Deep learning frameworks like TensorFlow, Pytorch and Caffe
  • Good command of Python, C/C++, Java and MATLAB
  • Leadership and teamwork
  • Working with efficiency
  • Innovative ideas
  • Fast learning skills