Hong Kong University of Science and Technology (HKUST)
Supervisor: Prof. Song Guo (IEEE Fellow, Chair Professor)
Topic: Multi-modal AI, Post-Training, and AI for Weather
> identity.profile
Hao Chen
PhD Candidate · Computer Science · Multi-modal AI
I am a PhD candidate at the Hong Kong University of Science and Technology (HKUST), supervised by Prof. Song Guo (IEEE Fellow, Chair Professor). I previously obtained my M.S. from Zhejiang University and B.S. from Shandong University. I am also affiliated with the Jindouyun (筋斗云) Program at ByteDance and Project Up (青云计划) at Youtu, Tencent.
10+Publications
CVPRHighlight '26
ICMLAccepted '26
ICCVAccepted '25
> ABOUT
My research focuses on Post-Training, On-Policy Distillation (OPD), Vision-Language Models, Mixture-of-Experts, Computer Vision, and AI for Weather. I serve as a reviewer for journals including TIP, TCSVT, TPAMI, and TOMM, as well as conferences including NeurIPS, ICLR, ICML, CVPR, ICCV, ECCV, ACM MM, AAAI, and IJCAI.
> RESEARCH_INTERESTS
> EDUCATION
Zhejiang University (ZJU)
Supervisor: Prof. Zhe-Ming Lu (IEEE Senior Member, Head of Department)
Topic: Computer Vision and Artificial Intelligence
Shandong University (SDU)
Rank: 2/21 · GPA: 89.7/100
State University of New York (SUNY)
Advisor: Prof. Zhongfei Zhang (IEEE Fellow)
Topic: Few-Shot Segmentation
University of Sheffield (UoS)
Advisor: Prof. Jungong Han (IAPR, AAIA Fellow)
Topic: Semantic Segmentation
> EXPERIENCE
Tencent · Youtu, Project Up Talent Program
Topic: Post-Training and On-Policy Distillation (Youtu-MM-Wiki)
ByteDance · Jindouyun Talent Program
Topic: Vision-Language Models and Post-Training
Shanghai AI Laboratory · Lei Bai Group
Topic: Mixture-of-Experts and AI for Weather
Huawei · 2012 Laboratory, Central Research Institute
Topic: Weather Forecasting (SKNO)
ByteDance · Data Department
Topic: Self-Supervised Visual Representation
Kuaishou · MMU, Vision Center
Topic: Self-Supervised Learning
> PUBLICATIONS
2026
Hao Chen, Tao Han, Jie Zhang, Song Guo, Fenghua Ling, and Lei Bai.
"EMFormer: Efficient Multi-Scale Transformer for Accumulative Context Weather Forecasting."
ICML, 2026.
CCF A
[pdf]
[code]
Hao Chen, Tao Han, Jie Zhang, Song Guo, and Lei Bai.
"STCast: Adaptive Boundary Alignment for Global and Regional Weather Forecasting."
CVPR, 2026.
Highlight
Top 2.5%
[pdf]
[code]
2025
Hao Chen, Tao Han, Song Guo, Jie Zhang, Yunlong Yu, Yonghan Dong, and Lei Bai.
"VA-MoE: Variables-Adaptive Mixture of Experts for Incremental Weather Forecasting."
ICCV, 2025.
CCF A
[pdf]
[code]
2024
Hao Chen, Yonghan Dong, Zheming Lu, Yunlong Yu, and Jungong Han.
"Pixel Matching Network for Cross-Domain Few-Shot Segmentation."
WACV, 2024.
Core A
[pdf]
[code]
2023
Hao Chen, Yonghan Dong, Zheming Lu, Yunlong Yu, and Jungong Han.
"Self-Prompting Perceptual Edge Learning for Dense Prediction."
IEEE TCSVT, 2023.
JCR Q1
[pdf]
[code]
Hao Chen, Yunlong Yu, Yonghan Dong, Zheming Lu, Yingming Li, and Zhongfei Zhang.
"Multi-Context Interaction Network for Few-Shot Segmentation."
ACM TOMM, 2023.
JCR Q1
[pdf]
[code]
Hao Chen, Yonghan Dong, Zheming Lu, Yunlong Yu, Yingming Li, and Jungong Han.
"Dense Affinity Matching for Few-Shot Segmentation."
Neurocomputing, 2023.
JCR Q1
[pdf]
[code]
Hao Chen, Zheming Lu, and Yangming Zheng.
"Multi-similarity Enhancement Network for Few-Shot Segmentation."
IEEE Access, 2023.
[pdf]
Hao Chen, Zhe-Ming Lu, and Jie Liu.
"A Monkey Swing Counting Algorithm Based on Object Detection."
IEICE Trans. Information and Systems, 2023.
[pdf]
Hao Chen and Zhe-ming Lu.
"Dynamic smoke detection by eliminating static targets in video."
IJICIC, 2023.
[pdf]
2022
> PROJECTS
AI for Weather · MoE Systems · 2024.09–2026.01
Built variable-adaptive dense MoE and sparse MoE acceleration for multi-variable weather forecasting at Shanghai AI Laboratory (+11.3% accuracy / −42% latency). Outcomes: VA-MoE, STCast, EMFormer. Earlier open-sourced SKNO at Huawei for global ERA5 forecasting.
Post-Training & Vision-Language Models · 2026.02–Present
Developing VLM post-training pipelines at ByteDance (Jindouyun) and Tencent Youtu (Project Up), covering SFT, instruction tuning, and on-policy distillation for risk control and multi-turn reasoning. Previously led long-text SFT at iFlytek with a divide-and-conquer summarization pipeline.
Few-Shot Segmentation · 2022.04–2023.06
MCINet reaches SOTA on 8 PASCAL/COCO benchmarks by fusing a learnable branch with a frozen backbone (ACM TOMM). Lightweight DAM / PMNet (0.68M params) handle cross-class / cross-dataset / cross-domain settings (WACV 2024, Neurocomputing).
Self-Prompt Learning · 2023.04–2023.10
SPPEL derives perceptual prompts from the input itself for dense prediction, without external prompt tokens (IEEE TCSVT).
> HONORS_AWARDS
National Encouragement Scholarship ×3 · 国家励志奖学金
Awarded the National Encouragement Scholarship for three consecutive years at Shandong University.
Outstanding Graduate of Shandong University · 山东大学优秀毕业生
Outstanding Graduate of Zhejiang Province · 浙江省优秀毕业生
Outstanding Graduate of Zhejiang University · 浙江大学优秀毕业生
ZJU Graduate Scholarship · 浙江大学毕业生奖学金
浙江大学毕业生奖学金（硕士+博士共1人）
Huawei Starlight Award · 华为公司星光奖
NOTE: This is a personal webpage. Views expressed here are solely those of the author and do not represent HKUST, ZJU, ByteDance, Tencent, or other affiliated organizations.