Hung-Chieh Fang

I received my bachelor’s degree in Computer Science from National Taiwan University, where I was fortunate to be advised by Professors Hsuan-Tien Lin, Shao-Hua Sun, and Yun-Nung (Vivian) Chen.

I was also a visiting research intern at Stanford University, working with Prof. Dorsa Sadigh, and at The Chinese University of Hong Kong, where I worked with Prof. Irwin King.

Email / Google Scholar / Github / X

News

- [June 2025] Our paper on improving generalization under non-IID data has been accepted to ICCV 2025. See you in Hawaii!

- [June 2025] I'm fortunate to visit ILIAD at Stanford, hosted by Prof. Dorsa Sadigh.

- [May 2025] Our paper on universal domain adaptation across any class priors has been accepted to ICML 2025. See you in Vancouver!

Research

I am interested in learning structured representations for generalist robotics. My research focuses on leveraging self-supervised learning to extract richer supervisory signals from data for policy learning and world modeling, with the goal of enabling generalization and continual learning.

All Robot Learning Representation Learning

DexDrummer: In-Hand, Contact-Rich, and Long-Horizon Dexterous Robot Drumming
Hung-Chieh Fang, Amber Xie, Jennifer Grannen, Kenneth Llontop, Dorsa Sadigh
Under Review
paper / website / code

We propose drumming as a unified testbed for in-hand, contact-rich, and long-horizon dexterous manipulation. We present DexDrummer, a hierarchical object-centric bimanual drumming policy trained in simulation with sim-to-real transfer.

Learning Skills from Action-Free Videos
Hung-Chieh Fang*, Kuo-Han Hung*, Chu-Rong Chen, Po-Jung Chou, Chun-Kai Yang, Po-Chen Ko, Yu-Chiang Wang, Yueh-Hua Wu, Min-Hung Chen, Shao-Hua Sun
ICML Workshop on Building Physically Plausible World Models, 2025
paper / website

We propose SOF, a method that leverages temporal structures in videos while enabling easier translation to low-level control. SOF learns a latent skill space through optical flow representations that better aligns video and action dynamics, thereby improving long-horizon performance.

	Soft Separation and Distillation: Toward Global Uniformity in Federated Unsupervised Learning Hung-Chieh Fang, Hsuan-Tien Lin, Irwin King, Yifei Zhang International Conference on Computer Vision (ICCV), 2025 paper / website / code / poster We explore how to improve generalization under highly non-IID data distributions where representations are non-shared. We propose a plug-and-play regularizer that encourages dispersion to improve uniformity without sacrificing semantic alignment.
	Tackling Dimensional Collapse toward Comprehensive Universal Domain Adaptation Hung-Chieh Fang, Po-Yi Lu, Hsuan-Tien Lin International Conference on Machine Learning (ICML), 2025 paper / website / poster We study how to adapt to arbitrary target domains without assuming any class-set priors. Existing methods suffer from severe negative transfer under large class-set shifts due to the overestimation of importance weights. We propose a simple uniformity loss that increases the entropy of target representations and improves performance across all class-set priors.
	Open-domain Conversational Question Answering with Historical Answers Hung-Chieh Fang, Kuo-Han Hung, Chao-Wei Huang, Yun-Nung Chen Asian Chapter of the Association for Computational Linguistics (AACL), 2022 paper / code We propose combining the signal from historical answers with the noise-reduction ability of knowledge distillation to improve information retrieval and question answering.

Awards

National Taiwan University

Principal’s Award for Bachelor’s Thesis, 2024 (Best Thesis in the EECS College)

Dean's List Award, Fall 2024 (Top 5% of the class)

Teaching and Service

	Reviewer ICLR (2026), TMLR (2025, 2026)
	National Taiwan University Teaching Assistant, EE5100: Introduction to Generative Artificial Intelligence, Spring 2024 Teaching Assistant, CSIE5043: Machine Learning, Spring 2023

This template is adapted from here.