Haosen Sun

Haosen Sun

Northwestern University

Welcome! I’m Haosen.

I will be starting as a Master’s student in Computer Science at Northwestern University in the fall of 2024. I am currently working as a research intern at Shanghai AI Lab. I completed my bachelor’s degree in Data Science and Technology at the Hong Kong University of Science and Technology, under the guidance of Prof. Chi-Keung Tang and Prof. Yu-Wing Tai. My research during my undergraduate studies focused on Neural Radiance Fields (NeRF), Diffusion Models (SD), and Neural Architecture Search (NAS). I also collaborated with Prof. Xiaomeng Li on Medical Vision Language Models (VLMs) as part of the UROP program.

I am passionate about several research areas, including Large Multimodal Models, Multi-modalities Generative AI, 3D Vision, and Efficient AI. My ultimate goal is to empower machines with the ability to extract meaningful patterns and relationships from both structured data (e.g. text, images, and video) and unstructured 3D geometric data. Also, I hope to enhance the interpretability and explainability of these models, propelling us closer to the realization of human-centric and physically-based general artificial intelligence.

I am seeking a PhD position starting in Fall 2025/26. Please reach out if you share similar research interests!

🔥 News

📝 Publications

* indicates equal contribution

ECCV 2024
sym

Auto-DAS: Automated Proxy Discovery for Training-free Distillation-aware Architecture Search

Haosen Sun, Peijie Dong, Zimian Wei, Shitong Shao, Lujun Li 

European Conference on Computer Vision (ECCV), 2024

[Project Code]

  • We present Auto-DAS, an automatic proxy discovery framework using an Evolutionary Algorithm (EA) for training-free Distillation-aware Architecture Search (DAS).
  • Auto-DAS generalizes well to various architectures and search spaces (e.g. ResNet, ViT, NAS-Bench-101, and NAS-Bench-201), achieving state-of-the-art results in both ranking correlation and final searched accuracy.
ECCV 2024
sym

Auto-GAS: Automated Proxy Discovery for Training-free Generative Architecture Search

Lujun Li, Haosen Sun, Shiwen Li, Peijie Dong, Qifeng Liu, Wei Xue, Yike Guo

European Conference on Computer Vision (ECCV), 2024

[Project Code]

  • We introduce Auto-GAS, the first training-free Generation Architecture Search (GAS) framework enabled by an auto-discovered proxy, which achieves competitive scores with 110× faster search than GAN Compression.
arXiv 2023
sym

Inpaint4DNeRF: Promptable Spatio-Temporal NeRF Inpainting with Generative Diffusion Models

Han Jiang*, Haosen Sun*, Ruoxuan Li*, Yu-Wing Tai, Chi-Keung Tang

Arxiv Preprint (Submitted to CVPR’24), Dec 2023

[Project Page] [Paper] [Project Code]

  • Inpaint4DNeRF can generate prompt-based objects guided by the seed images and their 3D proxies while preserving multiview consistency. Our generative baseline framework is general which can be readily extended to 4D dynamic NeRFs.
arXiv 2023
sym

Registering Neural Radiance Fields as 3D Density Images

Han Jiang*, Ruoxuan Li*, Haosen Sun, Yu-Wing Tai, Chi-Keung Tang

Arxiv Preprint, May 2023

[Paper]

  • We proposes a method to align and merge pre-trained NeRF models of partially overlapping 3D scenes using a generalized registration pipeline, incorporating key point detection, point set registration, and universal pre-trained descriptor networks with contrastive learning strategy.

Additional Publications

🎖 Honors and Awards

📖 Educations

  • 2024.09 - 2026.06 (now), M.S. in Computer Science, Northwestern University, USA
  • 2020.09 - 2024.07, BSc in Data Science and Technology, Hong Kong University of Science and Technology (HKUST), Hong Kong

💻 Internships

  • 08/2024 – present, Shanghai Artificial Intelligence Laboratory, China.

    Research Intern, working closely with Dr. Peng Ye.

  • 10/2023 – 05/2024, Hong Kong Generative AI Research and Development Center (HKGAI), Hong Kong.

    Research Intern, working closely with Dr. Lujun Li.