Hanchen Wang

Hey, I am a 2nd Year PhD (2019-) in Info Engineering (i.e. Computer Science) at Cambridge, where I work with Joan Lasenby and Adrian Weller, mainly on Efficient Geometric Deep Learning (i.e., 3D & Graph Representations).
Previously I worked with Hao Dong, Ali Javey and Xinran Wang on various topics incl. Electronic/Photovoltaic Devices and Computational Biology.
My PhD is supported via donations from Kathy Xu and Cambridge Trust.

I've obtained bachelor degrees in Physics and Info Engineering from Nanjing University (-2018, admitted at 15). During college, I've spent time in places incl. Berkeley EECS, Stanford EE, Meiji Japanese, MingShi Investment and China International Capital Corporation.

Email  |  Google Scholar  |  GitHub  |  LinkedIn  |  CV (Updated: Feb'21)

profile photo
Representative Papers [Full List]

'*' denotes equal contribution while '#' represents correspondence.

Advancing COVID-19 diagnosis with global privacy-preserving collaboration in artificial intelligence
Xiang Bai* Hanchen Wang*, Liya Ma*, Yongchao Xu*, Jiefeng Gan*, ..., Carola B. Schönlieb#, Tian Xia#
in review for Nature Machine Intelligence, 2021
an outdated version at medRxiv: 2020.05.10.20096073
main results  |  project page  |  code (Federated Learning , 3D DenseNets, CycleGAN on CT)
Unsupervised Point Cloud Pre-Training via View-Point Occlusion, Completion
Hanchen Wang, Qi Liu, Xiangyu Yue, Joan Lasenby, Matthew J. Kusner
in review for ICCV 2021
also presented at WeaSul ICLR 2021
project page  |  code  |  bibtex
Negative capacitance 2D MoS2 transistors with sub-60mV/dec subthreshold swing over 6 orders, 250 μA/μm current density, and nearly-hysteresis-free
Z. Yu*, H. Wang*, W. Li, R. Xu, X. Feng, P. Wang, P. Zhou, Y. Shi, Y. Chai, X. Wang#
IEDM, 2017   (Oral)
Open Source Contribution
deeplearning.ai: give internal feedbacks on the newly publicized courses on Natural Language Processing and TensorFlow2.

metric-learn: As part of scikit-learn-contrib, this module contains efficient Python implementations of supervised and weakly-supervised Mahalanobis metric learning algorithms.

pyntcloud: a Python library for working with 3D point clouds.

Teaching Assistant, Demonstrator and Grader
- 3F3: Statistical Signal Processing, Michaelmas 2019 and 2020
- 3F8: Statistical Inference, Lent 2020 and 2021
- Lab Part IB I1: Spectrum Analysis, Michaelmas 2020

- ECCV' 20, NeurIPS' 20, CVPR' 21, ACM TALLIP

- Puria Radmard (InfoEng, Trinity' 21, "Generative Models on Satellite images")

Talks and Pres
Research, Startup,
& Trading
- Quite a bite on some QishiCPC (Premium Member) seminars                                                 Since 2018

- Modeling Protein with Transformers @ Cambridge Genetics                                                      May 2021
- Neural Volume Rendering @ SigProc, Cambridge                                                                               Mar 2021
- A Collaborative Online AI Engine for CT-based COVID-19 Diagnosis @ CMIH                Aug 2020
- Machine Learning on Sets @ MLG, Cambridge                                                                                       Apr 2020
- 3D Deep Learning Meets Points Cloud @ St Edmund's College, Cambridge                        Feb 2020
- Federated ML with Differential Privacy @ CUED HackBridge Demo Day                            Oct 2019
- Acceptance Speech (Awardee Representative) @ Xuxin Fellowship Awarding Event   July 2018
- Commencement Speech (Student Representative) @ NJU KYM School Graduation      Jun 2018

- Read books, write notes and solve LeetCode puzzles
- Vice President of Cambridge Algorithmic Trading Society
- Know a whit about Classical Chinese, Japanese and French
- Achieved 3rd Level (entry level) of National Athlete in 400 metre dash
- Won 2nd Prize in China High School Biology Olympiad and Physics Olympiad
- Used to play Erhu (Chinese musical instrument, achieved 10th level (highest in amateur) at 11)

CSS is adapted from Jon Barron's, last modified: 2021-03