Huan Wang
Welcome to my homepage!👋🏻 My name is Huan Wang (Chinese: 王欢. My first name means "happiness or
joy" in Chinese). I am a Tenure-Track Assistant Professor at the AI Department of Westlake University(map-in-Chinese). I
did
my
Ph.D.
(2024) at Northeastern University (Boston, USA), advised by Prof. Yun
(Raymond) Fu at SMILE Lab. Before that,
I did my M.S. and B.E. at Zhejiang University (Hangzhou, China), advised
by Prof. Haoji Hu. I also visited VLLab at University of
California, Merced, luckily working with Prof. Ming-Hsuan Yang. I was fortunate to
intern at Google / Snap / MERL / Alibaba, working with many fantastic industrial researchers.
My research orbits Efficient AI in
vision and language modeling,
spanning image classifcation / detection / segmentation [GReg, PaI-Survey, TPP] to neural style transfer [Ultra-Resolution-NST],
single image super-resolution [ASSL/GASSL, SRP, ISS-P, Oracle-Pruning-Sanity-Check], 3D
novel
view synthesis / neural rendering / NeRF / NeLF [R2L,
MobileR2L, LightAvatar], AIGC /
diffusion models / Stable Diffusion [SnapFusion, FreeBlend], LLM /
MLLM [DyCoke, Poison-as-Cure], and snapshot
compressive imaging
(SCI) [QuantizedSCI, MobileSCI].
I also spent some time on easily
reproducible research.
My email: wanghuan [at] westlake [dot] edu [dot] cn
[ Google Scholar ]
[ GitHub ]
[ LinkedIn ]
[ Twitter ]
|
|
Openings / Collaborations
-
🔥Hiring Ph.D. Students / Research Assistants / Visiting Students
: I am actively looking for Ph.D. students (2026 Fall), research assistants, and
visiting students for research-oriented projects on efficient deep learning (e.g., network
pruning, distillation, quantization), GenAI (e.g., T2I generation, diffusion models, LLM/MLLM),
digital human / neural rendering, low-level vision (image / video restoration), etc.
Welcome to check here (and 知乎) for more information if you are
interested! Please fill this form when you
send me an email. Thanks!
|
Recent News
- 2025/02 [CVPR'25] DyCoke is accepted by CVPR'25!
Congrats to Keda!🎉
DyCoke is a training-free, plug-and-play token compression method for fast video LLMs: 1.5x
wall-clock inference speedup and 1.4x memory reduction with no performance drop. [arxiv][code]
- 2025/02 [Preprint] Can diffusion models blend visual
concepts that are semantically very unsimilar (e.g., an orange and a
teddy bear)? Yes, we introduce FreeBlend, a new method to blend
arbitrary concepts. [arxiv] [code] [webpage]
- 2025/01 [Preprint] Adversarial visual noise is always
malicious to our models like "poison"? No, we find it can also be a cure to mitigate the
hallucination problem of VLMs. [arxiv] [code] [webpage]
- 2025/01 [ICLR'25] One paper about distilling large
foundation models with low cost "Compressing Vision Foundation Models at ImageNet-level
Costs" is accepted by ICLR'25.
Thanks to the lead author Yitian!
- 2024/12 [Preprint] We present empirical evidence
to show that oracle pruning, the "ground-truth" pruning paradigm that has been followed for
around 35 years in the pruning community, does not hold in practice.
[arxiv][webpage]
- 2024/09 [NeurIPS'24] We introduce a training framework
Scala to learn slimmable ViTs. Using Scala, a ViT model is trained once but can inference
at different widths, up to the need of devices with different resources. The project is led by
Yitian. Congrats!
- 2024/07 [MM'24] We present the first real-time
on-device video SCI (Snapshot Compressive Imaging) framework via dedicated network design
and a distillation-based training strategy. Congrats to Miao!
- 2024/07 [ECCV'24] One paper about efficient video SCI
(Snapshot Compressive Imaging) via network quantization is accepted by ECCV'24 as an
oral. Congrats to Miao! [code]
- 2024/06 [New Start] Join the beautiful Westlake
University as a tenure-track assistant professor.
- 2024/04 [Graduation] 🎈PhD student→PhD. So many
thanks to my Ph.D. committee (Prof. Yun Raymond
Fu, Prof. Octavia
Camps, Prof.
Zhiqiang Tao) and co-authors!
- 2024/01 [ICLR'24] One paper proposing a simple and
effective data augmentation method for improved action recognition is accepted by ICLR'24.
Congrats to Yitian!
- 2023/11 [Award] Received Northeastern PhD Network
Travel Award. Thanks to Northeastern!
- 2023/10 [Award] Received NeurIPS'23 Scholar Award.
Thanks to NeurIPS!
- 2023/09 [NeurIPS'23] Two papers accepted by NeurIPS
2023: SnapFusion and a work about
weakly-supervised latent graph inference. 2024/02: Gave a talk at ASU about SnapFusion
[Slides]. Thanks for the warm invitation from Maitreya Jitendra Patel, Sheng Cheng, and Prof. Yezhou Yang!
- 2023/07 [ICCV'23] One paper about sparse network training for efficient
image super-resolution is accepted by ICCV'23. Congrats to Jiamian!
- 2023/06 [Internship'23] Start full-time summer
internship at Google AR.
- 2023/06 [Preprint] We are
excited to present SnapFusion,
a brand-new efficient Stable Diffusion model on mobile devices. SnapFusion can generate a
512x512 image from text in only 1.84s on iPhone 14 Pro, which is the
fastest🚀 on-device SD model as far as we know! Give a talk at JD Health about this
work [Slides].
- 2023/05 [Talk] Give a talk at ZJUI (ZJU-UIUC Institute) on the recent advances
in efficient neural light field (NeLF), featuring two of our recent NeLF papers (R2L and MobileR2L). [Slides]
- 2023/05 [Award] Recognized as CVPR'23 Outstanding
Reviewer (3.3%). Thanks to CVPR and the ACs!
- 2023/04 [TPAMI'23] Extension of our NeurIPS'21
spotlight paper
ASSL, "Global Aligned Structured Sparsity
Learning for Efficient Image Super-Resolution", is accepted by TPAMI (IF=24.31). Code
will be relased to the
same repo.
- 2023/03 [Award] Received ICLR'23 Travel Award. Thanks
to ICLR!
- 2023/02 [CVPR'23] 2 papers accepted by CVPR'23: (1) MobileR2L [Code]
, congrats to Junli! MobileR2L is a
blazing fast🚀 neural rendering model designed for mobile devices: It can render
1008x756 images at 56fps on iPhone13. (2) Frame Flexible
Network [Code], congrats to Yitian!
- 2023/01 [ICLR'23] 2 papers accpeted by ICLR'23: Trainabaility Preserveing Neural
Pruning (TPP) and Image as Set of
Points
(Oral, top-5%).
- 2023/01 [Internship'23] Start part-time internship at
Snap, luckily working with
the Creative Vision
team again. 2023/09 Update: The paper of this internship, SnapFusion, is accepted by NeurIPS
2023! Thanks to my coauthors!🎉
- 2023/01 [Preprint] Check out our preprint work that
deciphers the so confusing
benchmark
situation in neural network (filter) pruning: Why is
the State of Neural Network Pruning so Confusing? On the Fairness, Comparison Setup, and
Trainability in Network Pruning
[Code]. Also
give a talk @UT Austin about this work. Thanks for the warm invitation from Dr. Shiwei Liu and Prof. Atlas Wang!
- 2022/10 [Award] Received NeurIPS'22 Scholar Award.
Thanks to NeurIPS!
- 2022/09 [NeurIPS'22] 3 papers accepted by NeurIPS'22:
One under my lead (which was my 1st
internship work at MERL in 2020 summer. Rejected 4 times. Now finally I close the loop. Thanks
to
my co-authors and the reviewers!), two collaborations. Code: Good-DA-in-KD, PEMN, AFNet.
- 2022/09 [TIP'22] One journal paper "Semi-supervised Domain Adaptive
Structure Learning" accepted by TIP. Congtrats to Can!
- 2022/07 [ECCV'22] We present the first residual MLP
network to represent neural
light
field
(NeLF) for efficient novel view synthesis. Check our webpage and ArXiv!
- 2022/05 [Award] Received Northeastern PhD Network
Travel Award. Thanks to Northeastern!
- 2022/04 [IJCAI'22] We offer the very first survey paper
on Pruning at
Initialization,
accepted by IJCAI'22 [ArXiv] [Paper Collection]
[Slides].
- 2022/01 [ICLR'22] Two papers on neural network sparsity
accepted by ICLR'22. One is
about
efficient image super-resolution (SRP),
the other about lottery ticket hypothsis (DLTH).
- 2021/09 [NeurIPS'21] One paper on efficient image
super-resolution is accepted by
NeurIPS'21
as a Spotlight (<3%) paper! [Code]
- 2021/06 [Internship'21] Start summer internship at Snap
Inc., working with the
fantastic Creative
Vision team.
- 2021/01 [ICLR'21] One paper about neural network
pruning accepted by ICLR'21 as
poster. [ArXiv] [Code]
- 2020/06 [Internship'20] Start summer internship at
MERL, working with Dr. Mike
Jones and
Dr.
Suhas Lohit. (2022/09 Update: Finally,
paper of this
project accpeted by NeurIPS'22 -- two good years have passed, thank God..!)
- 2020/02 [CVPR'20] One paper about model compression for
ultra-resolution neural
style
transfer "Collaborative
Distillation for Ultra-Resolution Universal Style Transfer" is accepted by CVPR'20 [Code].
- 2020/01 [MLSys'20] 2019 summer intern paper accepted by
MLSys
2020. (Project: MNN from Alibaba, one of
the
fastest mobile AI engines on this planet. Welcome trying!)
- 2019/12 [JSTSP'19] One journal paper "Structured Pruning for Efficient
Convolutional Neural Networks via Incremental Regularization" accepted by IEEE
JSTSP.
- 2019/09 Join SMILE Lab at Northeastern University
(Boston, USA) to pursue my Ph.D. degree.
- 2019/07 [Internship'19] Start summer internship at
Taobao of Alibaba Group (
Hangzhou,
China).
- 2019/06 Graduate with M.Sc. degree from Zhejiang University
(Hangzhou, China).
|
|
DyCoke: Dynamic Compression of Tokens for Fast Video
Large Language Models
Keda Tao,
Can Qin,
Haoxuan You,
Yang Sui,
Huan Wang
In CVPR, 2025 |
ArXiv |
Code
|
|
Accessing Vision Foundation Models via
ImageNet-1K
Yitian Zhang,
Xu Ma,
Yue Bai,
Huan Wang,
Yun Fu
In ICLR, 2025 |
ArXiv |
Code
|
|
Is Oracle Pruning the True Oracle?
Sicheng Feng,
Keda Tao,
Huan Wang
Preprint, 2024 |
Webpage |
ArXiv |
Code
|
|
Towards Real-time Video Compressive Sensing on Mobile
Devices
Miao Cao, Lishun
Wang, Huan Wang, Guoqing Wang, Xin
Yuan
In ACM MM, 2024 |
ArXiv |
Code
|
|
A Simple Low-bit Quantization Framework for Video Snapshot
Compressive Imaging
Miao Cao,
Lishun Wang, Huan Wang, Xin
Yuan
In ECCV (Oral, 200/8585=2.3%), 2024 |
Arxiv |
Code
|
|
Don't Judge by the Look: A Motion Coherent Augmentation
for Video Recognition
Yitian Zhang, Yue Bai,
Huan Wang, Yizhou Wang, Yun Fu
In ICLR, 2024 |
ArXiv |
Code
|
|
SnapFusion: Text-to-Image Diffusion Model on Mobile
Devices within Two Seconds
Yanyu Li*,
Huan Wang*,
Qing Jin*,
Ju Hu,
Pavlo Chemerys,
Yun Fu,
Yanzhi Wang,
Sergey Tulyakov,
Jian Ren* (*Equal Contribution)
In NeurIPS, 2023 |
ArXiv |
Webpage
|
|
Iterative Soft Shrinkage Learning for Efficient Image
Super-Resolution
Jiamian
Wang, Huan Wang, Yulun
Zhang, Yun
Fu, Zhiqiang Tao
In ICCV, 2023 |
Arxiv |
Code
|
|
Why is the State of
Neural Network Pruning so Confusing? On the Fairness, Comparison Setup, and Trainability in
Network Pruning
Huan Wang, Can Qin, Yue Bai, Yun Fu
Preprint, 2023 |
ArXiv |
Code
|
|
Global Aligned Structured Sparsity
Learning for Efficient Image Super-Resolution
Huan Wang*, Yulun
Zhang*, Can Qin, Luc Van Gool, Yun Fu
(*Equal Contribution)
TPAMI, 2023 |
PDF |
Code
|
|
Real-Time Neural Light Field on Mobile
Devices
Junli Cao, Huan Wang, Pavlo
Chemerys, Vladislav Shakhray, Ju
Hu, Yun
Fu, Denys Makoviichuk,
Sergey Tulyakov, Jian Ren
In CVPR, 2023 |
Webpage |
ArXiv |
Code
|
|
Trainability Preserving Neural Pruning
Huan Wang, Yun
Fu
In ICLR, 2023 |
OpenReview |
ArXiv |
Code
|
|
Image as Set of Points
Xu Ma,
Yuqian Zhou,
Huan Wang, Can Qin, Bin
Sun, Chang
Liu, Yun
Fu
In ICLR (Oral, 5%), 2023 |
Webpage |
OpenReview |
Code
|
|
What Makes a "Good" Data Augmentation in
Knowledge Distillation -- A Statistical Perspective
Huan Wang, Suhas Lohit, Mike Jones, Yun
Fu
In NeurIPS, 2022 |
Webpage |
ArXiv |
Code
|
|
Parameter-Efficient Masking Networks
Yue Bai, Huan Wang, Xu
Ma, Yitian Zhang, Zhiqiang Tao, Yun
Fu
In NeurIPS, 2022 |
ArXiv |
Code
|
|
Look More but Care Less in Video Recognition
Yitian Zhang, Yue Bai, Huan
Wang, Yi Xu, Yun Fu
In NeurIPS, 2022 |
ArXiv |
Code
|
|
R2L: Distilling Neural Radiance Field to Neural Light
Field for Efficient Novel View
Synthesis
Huan Wang, Jian
Ren, Zeng Huang,
Kyle Olszewski, Menglei Chai, Yun Fu, Sergey Tulyakov
In ECCV, 2022 |
Webpage |
ArXiv |
Code
|
|
Recent Advances on Neural Network Pruning at
Initialization
Huan Wang, Can Qin, Yue Bai, Yulun Zhang, Yun Fu
In IJCAI, 2022 |
ArXiv | Slides
Paper
Collection
|
|
Learning Efficient Image Super-Resolution
Networks via Structure-Regularized Pruning
Huan Wang*, Yulun
Zhang*, Can Qin, Yun Fu
(*Equal Contribution)
In ICLR, 2022 |
PDF |
Code
|
|
Dual Lottery Ticket Hypothesis
Yue Bai, Huan
Wang, Zhiqiang Tao, Kunpeng
Li, Yun
Fu
In ICLR, 2022 |
OpenReview |
ArXiv |
Code
|
|
Aligned Structured Sparsity Learning for
Efficient Image Super-Resolution
Huan Wang*, Yulun
Zhang*, Can Qin, Yun Fu
(*Equal Contribution)
In NeurIPS (Spotlight, <3% ), 2021 |
PDF |
Code
|
|
Neural Pruning via Growing Regularization
Huan Wang, Can Qin, Yulun Zhang, Yun Fu
In ICLR, 2021 |
ArXiv,
OpenReview |
Code
|
|
Collabrotive Distillation for Ultra-Resolution Universal
Style Transfer
Huan Wang, Yijun
Li, Yuehai Wang, Haoji
Hu, Ming-Hsuan Yang
In CVPR, 2020 |
ArXiv, Camera
Ready |
Code
|
|
MNN: A Universal and Efficient Inference
Engine
Xiaotang Jiang, Huan Wang, Yiliu Chen, Ziqi Wu, et al.
In MLSys (Oral), 2020 |
ArXiv |
Code
|
|
Structured Pruning for Efficient ConvNets via Incremental
Regularization
Huan Wang, Xinyi Hu, Qiming Zhang, Yuehai
Wang, Lu
Yu, Haoji
Hu
In NeurIPS Workshop, 2018; IJCNN, 2019 (Oral);
Journal
extension to JSTSP, 2019
NeurIPS
Workshop,
IJCNN,
JSTSP |
Code
|
|
Structured Probabilistic Pruning for Convolutional Neural
Network Acceleration
Huan Wang*, Qiming Zhang*, Yuehai
Wang, Haoji
Hu (*Equal Contribution)
In BMVC (acceptance rate 29.5%), 2018 (Oral) |
ArXiv |
Camera Ready |
Code
|
Professional Services
- Journal Reviewer - TPAMI, IJCV, TIP, TNNLS, PR, IoT, JSTSP, TCSVT, etc.
- Conference Reviewer - CVPR/ECCV/ICCV/SIGGRAPH Asia, ICML/ICLR/NeurIPS, AAAI/IJCAI, COLM,
MLSys, etc.
|
|