Weisen JIANG (江伟森)

Postdoctoral Fellow
waysonkong@gmail.com
Department of Computer Science and Engineering
The Chinese University of Hong Kong (CUHK)

Research Interests: large language models, deep learning, machine learning
Google Scholar | OpenReview

🔥News

  • 2024 Dec: One paper (DCDM) was accepted by Neural Networks.
  • 2024 Sep: Two papers (RouterDC and GITA) were accepted by NeurIPS 2024.
  • 2024 Jul: Defended my Ph.D. thesis.
  • 2024 Jul: Two papers (MTMamba and MEHL-Soup) were accepted by ECCV 2024.
  • 2024 May: One paper (LETS-SAM) was accepted by ECML 2024.
  • 2024 May: One paper (FOBAR) was accepted by ACL 2024.
  • 2024 Jan: One paper (MetaMath) was accepted by ICLR 2024 (Spotlight).
News before 2024
  • 2023 Jan: One paper (AE-SAM) was accepted by ICLR 2023.
  • 2023 Apr: One paper (MetaPrompter) was accepted by ICML 2023.
  • 2022 May: One paper (MUSML) was accepted by ICML 2022.
  • 2021 Sep: One paper (MetaProx) was accepted by NeurIPS 2021.

Education

  • South China University of Technology, B.Eng. 2008 - 2012.

  • University of Chinese Academy of Sciences, M.Sc. (advisor: Prof. Hai-Tao Fang). 2012 - 2015.

  • The Hong Kong University of Science and Technology, Ph.D. (advisors: Prof. James Kwok and Prof. Yu Zhang). 2020 - 2024.

Work

  • Meituan, Engineer. 2015 - 2017.
    (i) recommendation systems and algorithms (ii) anti-cheating systems

  • Tencent, Senior Engineer. 2017 - 2019.
    (i) recommendation systems and algorithms (ii) big-data platforms

  • CUHK, Postdoctoral Fellow. 2024 - now.

Publications

(♠ equal contribution)

RouterDC: Query-Based Router by Dual Contrastive Learning for Assembling Large Language Models
Shuhao Chen♠, Weisen Jiang♠, Baijiong Lin, James T. Kwok, Yu Zhang
NeurIPS 2024, Vancouver, Canada
arXiv | Code | Poster

GITA: Graph to Visual and Textual Integration for Vision-Language Graph Reasoning
Yanbin Wei, Shuai Fu, Weisen Jiang, Zejian Zhang, Zhixiong Zeng, Qi Wu, James T. Kwok, Yu Zhang
NeurIPS 2024, Vancouver, Canada
arXiv | Project | Poster

MTMamba: Enhancing Multi-Task Dense Scene Understanding by Mamba-Based Decoders
Baijiong Lin, Weisen Jiang, Pengguang Chen, Yu Zhang, Shu Liu, Ying-Cong Chen
ECCV 2024, Milano, Italy
arXiv | Code | Poster | Follow-up work: MTMamba++

Learning Scalable Model Soup on a Single GPU: An Efficient Subspace Training Strategy
Tao Li♠, Weisen Jiang♠, Fanghui Liu, Xiaolin Huang, James Kwok
ECCV 2024, Milano, Italy
arXiv | Code | Poster

Forward-Backward Reasoning in Large Language Models for Mathematical Verification
Weisen Jiang, Han Shi, Longhui Yu, Zhengying Liu, Yu Zhang, Zhenguo Li, James Kwok
Findings of ACL 2024, Bangkok, Thailand
arXiv | Code | Project

MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models
Longhui Yu♠, Weisen Jiang♠, Han Shi, J. Yu, Z. Liu, Yu Zhang, James Kwok, Zhenguo Li, A. Weller, Weiyang Liu
ICLR 2024, Spotlight (acceptance rate 5%), Vienna, Austria
OpenReview |  Code |  Project |  Poster

Enhancing Sharpness-Aware Minimization by Learning Perturbation Radius
Xuehao Wang♠, Weisen Jiang♠, Shuai Fu, and Yu Zhang
ECML 2024, Vilnius, Lithuania
arXiv

Domain-Guided Conditional Diffusion Model for Unsupervised Domain Adaptation
Yulong Zhang, Shuhao Chen, Weisen Jiang, Yu Zhang, Jiangang Lu, James Kwok
Neural Networks, 2024
paper | code

Effective Structured Prompting by Meta-Learning and Representative Verbalizer
Weisen Jiang, Yu Zhang, James Kwok
ICML 2023, Honolulu, Hawaii, USA
OpenReview |  Code | Poster

An Adaptive Policy to Employ Sharpness-Aware Minimization
Weisen Jiang, Hansi Yang, Yu Zhang, James Kwok
ICLR 2023, Kigali, Rwanda
OpenReview |  Code | Poster

Subspace Learning for Effective Meta-Learning
Weisen Jiang, Yu Zhang, James Kwok
ICML 2022, Baltimore, Maryland, USA
OpenReview | Code | Poster

Effective Meta-Regularization by Kernelized Proximal Regularization
Weisen Jiang, James Kwok, Yu Zhang
NeurIPS 2021, Virtual
OpenReview | Code | Poster

SEEN: Few-Shot Classification with SElf-ENsemble
Weisen Jiang, James Kwok, Yu Zhang
IJCNN 2021, Virtual

Identification for Wiener System with Discontinuous Piece-wise Linear Function via Sparse Optimization
Weisen Jiang, Hai-Tao Fang
Chinese Control Conference (CCC) 2014

Patents

信息推荐方法、装置、系统及存储介质
江伟森, 吴德龙, 邱泰生
腾讯科技有限公司
发明专利, 授权号: CN110413868B, 2023

一种页面生成方法及装置
江伟森, 骆顺昌, 吴德龙, 邱泰生
腾讯科技有限公司
发明专利, 授权号: CN112307319A, 2021

Awards

Teaching

  • COMP 2012 (Object-oriented programming and data structures), Teaching assistant. HKUST.

  • COMP 4331 (Data mining), Teaching assistant. HKUST.

Academic Services

  • Journal Reviewer: TNNLS, AIJ
  • Conference Program Committee: ICML, NeurIPS, ICLR, CVPR, AAAI, AISTATS