Sizhe Lester Li

Sizhe Lester Li

李思哲

University of Rochester, MIT

I am a visiting student at MIT working on robot learning, hosted by Chuang Gan and Josh Tenenbaum.

I am also a final-year undergraduate in computer science at the University of Rochester, where I was fortunate to be advised by Chenliang Xu on audio-visual learning. Previously, I was fortunate to work with Thomas Serre on cancer detection, starting as a high school researcher at Brown University.

My research interests span robot learning, computer vision, and physics simulation. Currently, I develop methods for robots to learn to interact with deformable objects with challenging dynamics.

Email: sli96@u.rochester.edu, sizheli@mit.edu

Interests
  • Robot Learning: dynamics & control, differentiable physics
  • Computer Vision: scene representations
  • Computer Graphics: differentiable simulation
Education

  • Massachusetts Institute of Technology 2021 - Present

    Undergraduate Visiting Student

  • University of Rochester 2018 - Dec. 2022 (Exp.)

    B.S. in Computer Science

Research

Contact Points Discovery for Soft-Body Manipulations with Differentiable Physics
ICLR, 2022 (Spotlight, 5%)
Differentiable physics solver gets stuck when the contact points of the end effectors are sub-optimal. We propose a contact point discovery approach that guides the stand-alone differentiable physics solver to deform various soft-body plasticines.
 
PDF Cite Code Project
Contact Points Discovery for Soft-Body Manipulations with Differentiable Physics
Space-Time Memory Network for Sounding Object Localization in Videos
BMVC, 2021
Leveraging temporal synchronization and association within sight and sound is an essential step towards robust localization of sounding objects. To this end, we propose a space-time memory network for sounding object localization in videos.
 
Sizhe Lester Li*, Yapeng Tian*, Chenliang Xu
PDF Cite Code Project
Space-Time Memory Network for Sounding Object Localization in Videos
Learning to localize mutation in lung adenocarcinoma histopathology images
NeurIPS, 2020, LMRL Workshop
Molecular profiling of cancers is necessary to identify the optimal therapeutic options for patients. However, these assays are time-and-resource-intensive to perform, and they cannot accurately capture mutational heterogeneity. Here, we present a novel approach to address these issues.
 
Sahar Shahamatdar*, Daryoush Saeed-Vafa*, Drew Linsley*, Sizhe Lester Li, Sohini Ramachandran, Thomas Serre
PDF
Learning to localize mutation in lung adenocarcinoma histopathology images
A Deep Learning Algorithm for the Diagnosis and Gleason Grading of Whole Slide Images of Prostate Cancer Core Biopsies
The Journal of Urology, 2020
Training deep algorithms for diagnosis and grading of prostate cancer typically requires a large amount of manually annotated training data. Herein, we developed a weakly supervised approach for the diagnosis and gleason grading of prostate core biopsies.
 
Ohad Kott*, Sizhe Lester Li*, Drew Linsley, Ali Amin, Bora Golijanin, Dragan Golijanin, Thomas Serre
PDF
A Deep Learning Algorithm for the Diagnosis and Gleason Grading of Whole Slide Images of Prostate Cancer Core Biopsies

Teaching

Fall 2021: Deep Learning
Head TA
Sep 1, 2021
Fall 2021: Deep Learning
Spring 2020: Computer Vision
Teaching Assistant
Jan 1, 2020