Hanchen | Papers

My first-author papers (Nature, NeurIPS etc.) are highlighted by 1000+ tweets, 50+ press including:
A special collection of Nature, DeepMind, The Economist, Harvard Medical School, Chemical & Engineering News, Tech Crunch, Yahoo! News, a few Wikipedia pages, Tech Xplore, University of Cambridge, healthcare-in-europe.

Besides AI/ML, I'm trained and have published in Physics (my foundation and childhood's favorite, High School Olympiad), Biology (wild learner, also High School Olympiad), Chemistry, Materials / Electronics, Medical Sciences, and Clinical Trials :)

Postdoc, in Novel Therapeutics & Data-Centric AI for Science, Aug 2023-26

Along my postdoc fellowship, I'm excited about the potentials of single cell genomics and AI in developing novel therapeutics. I'm exploring the use of modern AI (e.g., Foundation Model, GenAI, ML System, LLM Agent) in single cell genomics areas including transcriptomics, imaging, spatial, perturbation, genetic variances & sequences modelling. Using Graph ML, I'm developing data-centric methods that could benefit many sciences (Biology, Math, Materials etc). Stay tuned!!

Fine-tuning Discrete Diffusion Models via Reward Optimization with Applications to DNA and Protein Design
Chenyu Wang*, Masatoshi Uehara*, et al., Tommi Jaakkola#, Sergey Levine#, Hanchen Wang#, Aviv Regev#
in review

Learning Multi-cellular Representations of scRNA Data Enables Characterization of Patient-level Disease States
Tianyu Liu*, Edward De Brouwer*, et al., Aviv Regev#, Graham Heimberg#
in review

Considerations for Building and Using Integrated Single-Cell Atlases
Karin Hrovatin*, Lisa Sikkema*, et al., Fabian Theis#, Malte Luecken#
Nature Methods 2024 (in press)

Metric Mirages in Cell Embeddings
Hanchen Wang, Jure Leskovec#, Aviv Regev#
Nature Biotechnology 2024 (in press), <1 month from submission to revision

PhD, in Machine Learning, Oct 2019-22

Doctoral Thesis: Learning from Structured Data with Weak Supervision

During my three-year PhD (with full fellowship), I shifted to AI, published my 1st paper till the end of my 2nd year (ICCV '21). I have been working on data including votes from polls, point clouds, CT/CXR scans, histological and pathological images, molecular and relational graphs, as well as scRNA-seq. I develop methods (many are pre-training) that can learn structures from these data with weak supervisions. Aside from this, I spent some time on quantum computation also did four interns in tech and biotech industries (Google -> Amazon -> BioMap -> Iambic). Back in 2019, I co-founded a startup (as the CTO) with friends from Stanford and Cal, aimed at transforming electronic health records via language models, with partnerships among 37 hospitals. Though things didn't work out then, my commitment to harnessing AI's power in healthcare remains steadfast.

Last batch of works are near completion: two cell atlases, a llm for math, a textbook on MLSys, a cell embedding model, a clinical trial on T-cell therapy. Stay tuned!!!

Unsupervised Discovery of Steerable Factors in Graphs
Shengchao Liu et al.,
TMLR 2024

Scientific Discovery in the Age of AI
Hanchen Wang et al., Connor Coley, Yoshua Bengio, Marinka Zitnik# [Team] >
Nature 2023, accept without revisions

Evaluating Self-supervised Learning for Molecular Graph Embeddings
Hanchen Wang*, Jean Kaddour*, Shengchao Liu, Jian Tang, Joan Lasenby, Qi Liu
NeurIPS 2023, Datasets and Benchmarks Track

Focalizing Regions of Relevance Facilitates Biomarker Prediction on Histopathological Images
Jiefeng Gan*, Hanchen Wang*, Hui Yu* et al., Tian Xia#
iScience 2023

Augmenting Message Passing by Retrieving Similar Graphs
Dingmin Wang et al.,
ECAI 2023

Machine Learning Systems: Design and Implementation
Tsinghua University 2023, Table of Contents
English version is in press, Springer Nature

Matching Point Sets with Quantum Circuits Learning
Hanchen Wang*, M. N.* (in contribution order)
ICASSP 2022, invited by editors, with travel award

Pre-training Molecular Graph Representation with 3D Geometry
Shengchao Liu, Hanchen Wang, Weiyang Liu, Joan Lasenby, Hongyu Guo, Jian Tang
ICLR 2022

Advancing COVID-19 Diagnosis with Privacy-Preserving Collaboration in AI
Xiang Bai (Prof.)* Hanchen Wang*, Liya Ma*, Yongchao Xu*, Jiefeng Gan* et al., Carola Schönlieb#, Tian Xia#
Nature Machine Intelligence 2021

Unsupervised Point Cloud Pre-training via Occlusion Completion
Hanchen Wang, Qi Liu, Xiangyu Yue, Joan Lasenby, Matthew J. Kusner
ICCV 2021

Iterative Teaching by Label Synthesis
Weiyang Liu*, Zhen Liu*, Hanchen Wang*, Liam Paul, Bernhard Schölkopf, Adrian Weller
NeurIPS 2021, Spotlight

Graph Denoising with Edge Editing
Hanchen Wang, Yunlong Jiao, Jordan Massiah
contributed talk at Amazon Machine Learning Conference 2021

Neural Random Subspace
Yun-Hao Cao et al.,
Pattern Recognition 2021

An Empirical Study on Learning Fairness Metrics for COMPAS Data with Human Supervision
Hanchen Wang, Nina Grgić-Hlača, Preethi Lahoti, Krishna. P. Gummadi, Adrian. V. Weller
axXiv, presented at NeurIPS HCML 2019, with travel award

Undergraduate, in Physics, Sep 2014-18

In addition to studying Physics, I researched next-gen electronic devices such as transistors and solar cells, an intersection of Solid-State Physics, Material Sciences and Electrical Engineering. I primarily worked in Xinran Wang's and Ali Javey's groups.

I also accumulated experience in the finance sector, initially focusing on bonds then transitioning to roles in quantitative trading of equity and crypto. These diverse experiences ultimately cemented my commitment in advancing Science and Engineering with the aim of improving human life. Finance is boring, technology is fun and the future :)

Current-controlled Propagation of Spin Waves in Antiparallel, Coupled Domains
Nature Nanotechnology 2019

Dopant‐Free Partial Rear Contacts Enabling 23% Silicon Solar Cells
Advanced Energy Materials 2019

Stable dopant-free asymmetric heterocontact silicon solar cells with efficiencies above 20%
ACS Energy Letters 2018

Negative Capacitance 2D MoS2 Transistors with Sub-60mV/dec Subthreshold Swing over 6 Orders, 250 μA/μm Current Density, and Nearly-Hysteresis-Free
Zhihao Yu (Ph.D)*, Hanchen Wang (B.S)* et al., Xinran Wang#
IEDM 2017, Oral, NJU's 1st IEDM. It is where Intel, NVIDIA, TSMC, AMD etc sharing their secret sauce on chip design :)

Molecular Mechanism of Self-Assembly of Aromatic Oligoamides into Interlocked Double-Helix Foldamers
Journal of Physical Chemistry B 2017

Microchannel contacting of crystalline silicon solar cells
Scientific Reports 2017

Logical integration device for two-dimensional semiconductor transition metal sulfide
Weisheng Li*, Jian Zhou*, Hanchen Wang* et al., Xinran Wang#
Invited Review, Acta Physica Sinica 2017