Tao Huang

Ph.D. Student
The University of Sydney

Biography

My name is Tao Huang (黄涛). I am a 2nd-year PhD student at The University of Sydney, advised by Prof. Chang Xu. Before that, I was a Researcher at SenseTime, supervised by Dr. Shan You. In 2020, I received my Bachelor's degree in Computer Science at Huazhong University of Science and Technology, China.

My major research interests lie within efficient deep learning algorithms, such as

Efficient architectures: NAS, pruning, Rep, handcraft
Knowledge distillation (KD)
Efficient training

I am opening for job opportunities and collaborations, please drop me an email if interested.

News

[2024/03] Our PU-MLC was accepted to ICME 2024 as oral.
[2024/03] We released ActGen & LocalMamba & EfficientVMamba at arXiv.
[2024/02] One paper (FreeKD) was accepted to CVPR 2024.
[2024/02] We released DIST+.
[2023/09] One paper (DiffKD) was accepted to NeurIPS 2023.
[2023/07] One paper (AKD) was accepted to ACM MM 2023.
[2023/05] We released DiffKD and AKD at arXiv.
[2023/01] One paper (MasKD) was accepted to ICLR 2023.
[2022/09] One paper (DIST) was accepted to NeurIPS 2022.

[2022/07] We released LightViT [code] at arXiv.
[2022/05] We released DIST [code] and MasKD [code] at arXiv.
[2022/04] One paper (QAS) was accepted to Nature Partner Journals Quantum Information (NPJ QI).
[2022/03] Two papers (GreedyNASv2 & DyRep) were accepted to CVPR 2022.
[2022/01] One paper (ReLoss) was accepted to ICLR 2022.
[2021/12] We released MMRazor - a model compression toolkit for model slimming and AutoML, which includes 3 mainstream technologies NAS, pruning, and KD. MMRazor can be easily applied to various projects (e.g., MMDet and MMCls) in OpenMMLab.
[2021/11] We released GreedyNASv2 at arXiv.
[2021/03] One paper about NAS was accepted to CVPR 2021. The NAS benchmark in our paper was released at github.
[2021/01] One paper about channel number search (pruning) was accepted to ICLR 2021 as spotlight.
[2020/11] One paper about NAS was released at arXiv. Our TopoNAS explicitly learns the topology for differentiable NAS (DARTS), and enjoys significant efficiency improvement on obtained architectures.
[2020/10] One paper about quantum architecture search (QAS) was released at arXiv. Our QAS implicitly learns a rule that can well suppress the influence of quantum noise and the barren plateau.
[2020/02] One paper about NAS was accepted to CVPR 2020.

Publications

First-authored papers: CVPRx 4, NeurIPSx 2, ICLRx 2 *: co-first author.

ICMEPositive Label Is All You Need for Multi-Label Classification🚩 Oral

Zhixiang Yuan, Kaixin Zhang, Tao Huang
IEEE International Conference on Multimedia and Expo (ICME), 2024.

ArXiv Bib
CVPRFreeKD: Knowledge Distillation via Semantic Frequency Prompt

Yuan Zhang, Tao Huang, Jiaming Liu, Tao Jiang, Kuan Cheng, Shanghang Zhang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

ArXiv Bib
NeurIPSKnowledge Diffusion for Distillation

Tao Huang, Yuan Zhang, Mingkai Zheng, Shan You, Fei Wang, Chen Qian, Chang Xu
Advances in Neural Information Processing Systems (NeurIPS), 2023.

ArXiv Bib Poster
ACM MMAvatar Knowledge Distillation: Self-ensemble Teacher Paradigm with Uncertainty

Yuan Zhang, Weihua Chen, Yichen Lu, Tao Huang, Xiuyu Sun, Jian Cao
Proceedings of the 31th ACM International Conference on Multimedia (ACM MM), 2023.

ArXiv Bib
ICLRMasked Distillation with Receptive Tokens

Tao Huang*, Yuan Zhang*, Shan You, Fei Wang, Chen Qian, Jian Cao, Chang Xu
International Conference on Learning Representations (ICLR), 2023.

ArXiv
NeurIPSKnowledge Distillation from A Stronger Teacher

Tao Huang, Shan You, Fei Wang, Chen Qian, Chang Xu
Advances in Neural Information Processing Systems (NeurIPS), 2022.

ArXiv Slides Poster 解读
NPJ QIQuantum circuit architecture search for variational quantum algorithms

Yuxuan Du, Tao Huang, Shan You, Min-Hsiu Hsieh, Dacheng Tao
Nature Partner Journals Quantum Information (NPJ QI), 2022.

ArXiv
CVPRGreedyNASv2: Greedier Search with a Greedy Path Filter

Tao Huang, Shan You, Fei Wang, Chen Qian, Changshui Zhang, Xiaogang Wang, Chang Xu
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022.

ArXiv Poster
CVPRDyRep: Bootstrapping Training with Dynamic Re-parameterization

Tao Huang, Shan You, Bohan Zhang, Yuxuan Du, Fei Wang, Chen Qian, Chang Xu
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022.

ArXiv Poster
ICLRRelational Surrogate Loss Learning

Tao Huang, Zekang Li, Hua Lu, Yong Shan, Shusheng Yang, Yang Feng, Fei Wang, Shan You, Chang Xu
International Conference on Learning Representations (ICLR), 2022.

ArXiv Bib Poster Slides
ICASSPData Agnostic Filter Gating for Efficient Deep Networks

Hongyan Xu, Xiu Su, Shan You, Tao Huang, Fei Wang, Chen Qian, Changshui Zhang, Chang Xu, Dadong Wang, Arcot Sowmya
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022.

ArXiv Bib
CVPRPrioritized Architecture Sampling with Monto-Carlo Tree Search

Xiu Su*, Tao Huang*, Yanxi Li, Shan You, Fei Wang, Chen Qian, Changshui Zhang, Chang Xu
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021.

ArXiv Poster Bib NAS-Bench-Macro
ICLRLocally Free Weight Sharing for Network Width Search🚩 Spotlight

Xiu Su, Shan You, Tao Huang, Fei Wang, Chen Qian, Changshui Zhang, Chang Xu
International Conference on Learning Representations (ICLR), 2021.

ArXiv Bib OpenReview
CVPRGreedyNAS: Towards Fast One-Shot NAS with Greedy Supernet

Shan You*, Tao Huang*, Mingmin Yang*, Fei Wang, Chen Qian, Changshui Zhang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2020.

ArXiv Bib Poster Video Slides

Manuscripts

arXivEfficientVMamba: Atrous Selective Scan for Light Weight Visual Mamba

Xiaohuan Pei*, Tao Huang*, Chang Xu
arXiv preprint arXiv:2403.09977 (2024).

ArXiv
arXivLocalMamba: Visual State Space Model with Windowed Selective Scan

Tao Huang, Xiaohuan Pei, Shan You, Fei Wang, Chen Qian, Chang Xu
arXiv preprint arXiv:2403.09338 (2024).

ArXiv
arXivActive Generation for Image Classification

Tao Huang, Jiaqi Liu, Shan You, Chang Xu
arXiv preprint arXiv:2403.06517 (2024).

ArXiv
arXivNot All Steps are Equal: Efficient Generation with Progressive Diffusion Models

Wenhao Li, Xiu Su, Shan You, Tao Huang, Fei Wang, Chen Qian, Chang Xu
arXiv preprint arXiv:2312.13307 (2023).

ArXiv
DIST+: Knowledge Distillation from A Stronger Teacher

Tao Huang, Shan You, Fei Wang, Chen Qian, Chang Xu

PDF
arXivLightViT: Towards Light-Weight Convolution-Free Vision Transformers

Tao Huang, Lang Huang, Shan You, Fei Wang, Chen Qian, Chang Xu
arXiv preprint arXiv:2207.05557 (2022).

ArXiv Bib
arXivExplicitly Learning Topology for Differentiable Neural Architecture Search

Tao Huang, Shan You, Yibo Yang, Zhuozhuo Tu, Fei Wang, Chen Qian, Changshui Zhang
arXiv preprint arXiv:2011.09300 (2020).

ArXiv

Grants

2023-2024, Efficient Language-and-Vision Models, 1000 USD, Google Cloud Research Credits Award

Talks

TMLR Young Scientist Seminar@HKBU: "Knowledge Distillation from A Stronger Teacher", Jul. 2022. [Slides]

Teaching

2024-2025, Guest Lecturer for COMP5329 Deep Learning, The University of Sydney
2024-2025, Tutor for COMP5329 Deep Learning, The University of Sydney

Academic Services

Reviewer for Conferences:

NeurIPS 2021-2023, ICLR 2023-2024, ICML 2022-2024, CVPR 2022-2024, ICCV 2023, ECCV 2024, AAAI 2023-2024, ACM MM 2021-2023.

Reviewer for Journals:

TNNLS, TIP, IJCV

Education

The University of Sydney, Sydney, Australia

Jul. 2022 – Jun. 2025 (expected)

Ph.D. student at School of Computer Science
Advisor: Prof. Chang Xu
Huazhong University of Science and Technology, Wuhan, China

Sep. 2016 – Jun. 2020

B.E. in Computer Science and Technology

Selected Experiences

SenseTime, Beijing, China

Research Intern

Jul. 2022 - Present

Researcher

Jul. 2020 – Jun. 2022

Research Intern

Aug. 2019 – Jul. 2020
- Research on model compression algorithms (NAS, KD, pruning, etc.).
- Research and development on auto-driving (smart carbin) scenes such as face verification and drowsiness detection.
- Applied NAS to face verification task on largescale industrial datasets, which significantly improves the performance.
Horizon Robotics, Beijing, China

Computer Vision Research Intern

May 2019 – Aug. 2019
- Development on object detection framework: anchor-free detection method, detection in traffic scene.
- Research on knowledge distillation methods for face alignment, object detection.

Dian Group, Wuhan, Hubei province, China

Team Leader of Realtime Face Detection & Alignment Project, AI Group

Nov. 2018 – May 2019
- Develop Android APP to inference face detection, tracking, and 106-point landmark models on mobile devices.
- Research on model acceleration (e.g., knowledge distillation, model pruning) and facial landmark (e.g., multitask learning, loss function, augmentation).
- Our proposed model archieves an inference speed of 5 ms / image on Huawei Mate20 Pro.
Core Member of Beibei Intelligent Customer Service Project, AI Group

Feb. 2018 – Nov. 2018
- This project comes from Beibei Group Company, Beibei is the bigest mother-baby ecommerce platform in China. The task is to find optimal answers based on classifications of customer questions.
- Research and development on text classification and data augmentation, etc.
3D Printer Team, Wuhan, Hubei province, China

Group Learder of Embedded Control Group

Oct. 2016 – Jan. 2019

Topic: developing control algorithms for 3DP/FDM 3D printers

Selected Awards

Outstanding Graduate award, Huazhong University of Science and Technology

Jun. 2020
Outstanding Graduate Thesis award, Huazhong University of Science and Technology

Jun. 2020