Weichen Huang

weichen-huang.github.io

wh@gatech.edu

weichen-huang

weichenhuang1

weichen-huang

Atlanta, Georgia 30332 USA

+1-943-235-5093

Interests

Mechanistic Interpretability , Representation Learning , Computational Neuroscience , Multimodal Learning

Skills

Programming: Python, PyTorch, C++

Research Areas: Mechanistic Interpretability, Representation Learning, Computer Vision, NLP, Computational Neuroscience

Education

Georgia Institute of Technology

B.S. in Computer Science (Class of 2029), Atlanta, GA

GPA: 4.0/4.0

2025 - Present

St Andrew's College

International Baccalaureate, Dublin, Ireland

GPA: 43/45; SAT 1580

  • HL: Math AA (7), Physics (7), Chemistry (7)

2019 - 2025

Work

Undergraduate Researcher, Georgia Tech - Language Intelligence Lab

www.language-intelligence-thought.net/

PI: Dr. Anna Ivanova

  • Interpreting LLM-based fMRI encoding models using mechanistic interpretability
  • Building fine-grained semantic maps of the human brain's language network using Representational Similarity Analysis

Oct 2025 - Present

RSI Researcher, MIT - DiCarlo Lab

dicarlolab.mit.edu/

PI: Dr. James J. DiCarlo

  • Investigated primate ventral visual stream representations for spatial latent estimation
  • ICLR 2025 paper: Vision Models Trained to Estimate Spatial Latents Learn Ventral-Stream-Aligned Representations

Jun 2024 - Oct 2024

Research Intern, University College Dublin - ML in Medical Imaging Lab

www.ucd-ml-mi.com/

PI: Dr. Kathleen Curran

  • Developed multimodal contrastive learning frameworks for medical diagnosis (ICCV 2023, NeurIPS 2024)
  • Built MRI-PET fusion and graph-based representation learning methods for Alzheimer's detection

Feb 2023 - Jan 2025

Activities

Rickoid, RSI-MIT 2024

www.cee.org/programs/research-science-institute

Research Science Institute at MIT (22 June - 3 August)

  • Presented 'Convergent Neural Representations' research at MIT BCS DiCarlo Lab

2024 - 2024

Reviewer, ICLR 2025

iclr.cc/

Conference Reviewer for ICLR 2025

2025 - 2025

Publications

Vision Models Trained to Estimate Spatial Latents Learn Ventral-Stream-Aligned Representations, Y. Xie, W. Huang, E. Alter, J. Schwartz, J. Tenenbaum, J. DiCarlo, , 2025

computing.mit.edu/news/a-visual-pathway-in-the-brain-may-do-more-than-recognize-objects/

ICLR 2025

Multimodal Representation Learning using Adaptive Graph Construction, Weichen Huang, , 2024

arxiv.org/abs/2410.06395

NeurIPS 2024 (High School Projects, top 4/335)

Multimodal Contrastive Learning and Tabular Attention for Automated Alzheimer's Disease Prediction, Weichen Huang, , 2023

arxiv.org/abs/2308.15469

ICCV 2023

Spurious Correlation Mitigation in CXR Images via Reinforcement Learning and Self-Supervision, Weichen Huang, K. Curran, , 2023

zenodo.org/records/8245236

IMVIP 2023

Feature Scaling and Attention Convolutions for Extreme Precipitation Prediction, Weichen Huang, , 2022

ai4good.org/fragile-earth-2022/

ACM KDD 2022 Workshop

Awards

NeurIPS'24 Paper Award, NeurIPS, 2024

High School Projects track (top 4/335)

IOI Irish National Team, IOI, 2025

International Olympiad in Informatics

USACO Gold, USACO, 2023

RSI 2024 Scholar, CEE/MIT, 2024

Research Science Institute at MIT

ICLR 2025 Reviewer, ICLR, 2025

BT Young Scientist Winner, BTYSE, 2021-2024

4x Category Winner (2021-2024); Irish Research Council Award (2024)

References

Available on request