Kaixiong Gong

CUHK MMLab. Hong Kong.

prof_pic.jpg

MMLab, The Chinese University of Hong Kong

Hong Kong

I am a Ph.D. student at MMLab, The Chinese University of Hong Kong, advised by Prof. Xiangyu Yue. Before this, I worked with Prof. Shuang Li and Prof. Chi Harold Liu at Beijing Institute of Technology.

My research interests lie in AR generative models for video generation and multi-modal language models. I am always open to discussion and collaboration — feel free to drop me an email.

News

Aug 05, 2026 2026: Twins: Learn to Predict Unified Representations with Focal Loss was accepted to ICML 2026; Probing Audio-Visual Reasoning in Multimodal Language Models through the Lens of Audio was accepted to ACL 2026 (Oral); Learning While Staying Curious: Entropy-Preserving Supervised Fine-Tuning via Adaptive Self-Distillation for Large Reasoning Models was accepted to ACL 2026.
Dec 31, 2025 2025: Video-R1: Reinforcing Video Reasoning in MLLMs was accepted to NeurIPS 2025.
Dec 31, 2024 2024: BIFRÖST: 3D-Aware Image Compositing with Language Instructions was accepted to NeurIPS 2024; OneLLM, Multimodal Pathway, and Text-to-3D Generation with Bidirectional Diffusion were accepted to CVPR 2024; Reg-TTA3D was accepted to ECCV 2024; Subjective Topic meets LLMs was accepted to EMNLP 2024.
Dec 31, 2023 2023: Meta-Transformer: A Unified Framework for Multimodal Learning and Plain Language Models Can Read Point Clouds and Benefit 3D Vision were released as preprints.
Dec 31, 2022 2022: Improving Transferability for Domain Adaptive Detection Transformers was accepted to ACM Multimedia 2022.
Dec 31, 2021 2021: Pareto Domain Adaptation was accepted to NeurIPS 2021; MetaSAug and Transferable Semantic Augmentation were accepted to CVPR 2021, with Transferable Semantic Augmentation selected as an Oral presentation.

Experience

Hunyuan @Tencent
10/2025 - Present
Shenzhen
Research Intern
Mentors: Jian-Wei Zhang and Miles Yang
Research Topics: Unified Understanding and Generation, Video Generation
Core Contribution: Hunyuan Video
Contribution: Hunyuan Image
Kuaishou Tech @Kling
05/2024 - 08/2025
Shenzhen
Research Intern
Mentor: Xintao Wang
Research Topics: Autoregressive Visual Generation, Unified Understanding and Generation
Shanghai AI Lab
11/2022 - 05/2024
Beijing
Research Intern
Mentor: Xiangyu Yue
Research Topics: Multi-modal LLM

Education

The Chinese University of Hong Kong
Sep. 2023 - Present
Hong Kong, China
Ph.D. in MMLAB (Information Engineering)
Honor: Hunyuan Scholarship
Supervisor: Prof. Xiangyu Yue
Research Topics: Multi-modal Learning, Generative Models
Beijing Institute of Technology
2020 - 2023
Beijing, China
Master's Degree in Computer Science
Honor: National Scholarship, Ministry of Education of China (2021)
Beijing Institute of Technology
2016 - 2020
Beijing, China
Bachelor's Degree in Software Engineering

Research

  1. ICML
    twins.png
    Twins: Learn to Predict Unified Representations with Focal Loss
    Kaixiong Gong, Xin Cai, Bin Lin, and 9 more authors
    In International Conference on Machine Learning (ICML), 2026
    * Equal contribution
    arXiv PDF Code
  2. ACM MM
    mm2022_detr.png
    Improving Transferability for Domain Adaptive Detection Transformers
    Kaixiong Gong, Shuang Li, Shugang Li, and 3 more authors
    In Proceedings of the 30th ACM International Conference on Multimedia (ACM MM), 2022
    PDF
  3. IEEE TCYB
    tcyb_CSDN.png
    Critical Classes and Samples Discovering for Partial Domain Adaptation
    Shuang Li, Kaixiong Gong, Binhui Xie, and 3 more authors
    IEEE Transactions on Cybernetics, 2022
    PDF
  4. NeurIPS
    nips2021_pareto.png
    Pareto Domain Adaptation
    Fangrui Lv, Jian Liang, Kaixiong Gong, and 5 more authors
    In Advances in Neural Information Processing Systems (NeurIPS), 2021
    PDF Code
  5. CVPR
    cvpr2021_metasaug.png
    MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition
    Shuang Li, Kaixiong Gong, Chi Harold Liu, and 3 more authors
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021
    PDF Code
  6. CVPR (Oral)
    cvpr2021_tsa.png
    Transferable Semantic Augmentation for Domain Adaptation
    Shuang Li, Mixue Xie, Kaixiong Gong, and 3 more authors
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021
    PDF Code
  7. IEEE TKDE
    tkde_ms.png
    End-to-End Transferable Anomaly Detection via Multi-spectral Cross-domain Representation Alignment
    Shuang Li, Shugang Li, Mixue Xie, and 4 more authors
    IEEE Transactions on Knowledge and Data Engineering (TKDE), 2021
    PDF

Services

Conference Reviewer
NeurIPS, ICML, CVPR, ICCV

Invited Talks

MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition
VALSE
10/2021
MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition
BAAI (Beijing Academy of Artificial Intelligence)
03/2021