Me @Tokyo, Japan

Thomas Huang

University of Michigan

Address: 2260 Hayward St., Ann Arbor, MI

Email: thomaseh [at] umich [dot] edu

About Me

Hello! My name is Thomas Huang. I am a first-year PhD student in the EECS department at the University of Michigan. My advisor is Professor Honglak Lee. My research interests are deep learning and its application to computer vision.

Click on the tabs for more details


Learning Hierarchical Semantic Image Manipulation through Structured Representations
Seunghoon Hong, Xinchen Yan, Thomas Huang, Honglak Lee
To appear in Advances in Neural Information Processing Systems (NeurIPS), Montreal, Canada, 2018
[arxiv] [code]

Research Experience

AI Lab – University of Michigan Apr. 2018 – current
Berkeley DeepDrive (BDD) Lab – University of California, Berkeley Jun. 2019 – current
Research Intern
ARM Lab – University of Michigan Jan. 2018 – Apr. 2018
Research Assistant


EECS 445: Intro. to Machine Learning – University of Michigan Jan. 2018 – Apr. 2019
Instructional Aide (2 semesters), Head Instructional Aide (1 semester)

I taught fundamental machine learning concepts, including linear classifiers, support vector machines, decision trees, bagging and boosting, neural networks, deep learning, clustering, collaborative filtering, and graphical models.


University of Michigan – Ann Arbor, Michigan Sep. 2019 – Apr. 2024 (expected)
Ph.D. in Computer Science and Engineering
University of Michigan – Ann Arbor, Michigan Sep. 2015 – Apr. 2019
Bachelor of Engineering in Computer Science and Data Science

Cumulative GPA: 3.99, Major GPA: 4.0
Relevant coursework: Machine Learning, Reinforcement Learning, Computer Vision, Artificial Intelligence, Natural Language Processing, Data Mining, Algorithmic Robotics, Web Systems, Operating Systems.

Honors and Awards: Dean's Honor List (Fall '15, Winter & Fall '16, Fall '17, Winter '18), University Honors (Winter '16, Fall '17, Winter '18), James B. Angell Scholar ('18)

Honors and Awards

  • NSF Graduate Research Fellowship Honorable Mention, 2018
  • Richard H. Orenstein Fellowship (in Memory of Murray Orenstein), 2019
    • Endowed in memory of Murray Orenstein in 1979 by Richard H. Orenstein (BS '62), Sarasota, Florida


ICLR 2018 Reproducibility Challenge Sep. 2017 – Dec. 2017
Final Project for EECS 498: Reinforcement Learning at the University of Michigan

For this challenge, we aimed to reproduce the results from the paper Curiosity-driven Exploration by Bootstrapping Features. Our implementation and report can be found here.
Front-end Collision Prevention System Sep. 2017 – Dec. 2017
Final Project for EECS 442: Computer Vision at the University of Michigan

My team created a system to be incorporated into dash cameras to automatically detect cars suddenly breaking/slowing down in front. For the detection of cars, we used the Single Shot MultiBox Detector (SSD) with pre-trained weights on the VOC2007 dataset. For detecting break lights, we used a sequence of HSL thresholding, morphological closure, and blob detection. A combination of both detections are used to determine whether the cars in front are suddenly breaking/slowing down. The results of our system can be found here and here. The color of the bounding box indicate the detection result. The final report can be found here.

Professional Experience

Salesforce – San Francisco, California May. 2017 – Aug. 2017
Frontier Scale Software Engineering Intern


UM::Autonomy – University of Michigan Sep. 2015 – Sep. 2018
Lead for the AI subteam (2016 – 2018)

Our team designs and builds an autonomous, robotic boat every year to compete in the International RoboBoat Competition, which is composed of several challenges. Our team also composes a research journal paper/technical design report each year, which can be found here (2016, 2017). For two years, I was the lead of the AI subteam, which is in charge of the software and algorithms used by the boat. I supervised various tasks, such as data collection, object detection, and control systems.
Eta Kappa Nu, EECS Honor Society – University of Michigan Sep. 2016 – Apr. 2019


NeurIPS 2018 – Montreal, Canada Dec 3, 2018 – Dec 8, 2018
Poster Session

I presented our paper, Learning Hierarchical Semantic Image Manipulation through Structured Representations, at the poster session (pictures).


An outdated list of papers that I read for research/projects, which can be found here.