Xuanyi Dong (董宣毅)
School of Computer Science, University of Technology Sydney (UTS)
Address: 15 Broadway, Ultimo NSW 2007, Australia
Email: xuanyi.dong [at] student.uts.edu.au

About Me ([GitHub] [Google Scholar] [Full Publications])

I am a third-year PhD candidate at School of Computer Science, University of Technology Sydney (UTS). My academic supervisor is Prof. Bogdan Gabrys. I received the bachelor's degree from School of Computer Science and Technology, Beihang University (BUAA) in 2016. My research interests included automated deep learning and its application to real world applications.

My research is supported by Google PhD Fellowship (2019).

Selected Industrial Experience

Invited Talks

  • Sep 2020: Efficient Differentiable Automated Deep Learning, UT Austin
  • Apr 2020: Towards Efficient and Reproducible Neural Architecture Search, Baidu USA, Mountain View
  • Summer 2019: Efficient NAS and its Applications to Computer Vision, YITUTech, Baidu, SCUT, SUST

Selected Publications

AutoHAS: Differentiable Hyper-Parameter and Architecture Search
Xuanyi Dong, Mingxing Tan, Adams Wei Yu, Daiyi Peng, Bogdan Gabrys, Quoc V. Le
in preprint, 2020
[ Paper ] [ BibTex ]
A differentiable hyper-parameter and architecture search approach, and we show improvements on MobileNet/ResNet/EfficientNet/BERT.
NATS-Bench: Benchmarking NAS algorithms for Architecture Topology and Size
Xuanyi Dong, Lu Liu, Katarzyna Musial, Bogdan Gabrys
an extended version of NAS-Bench-201, which is published in ICLR 2020.
Current version: [arXiv] [Code] [Website] [BibTex]
ICLR version: [PDF] [Code] [arXiv] [Package] [Website] [BibTex]
An algorithm-agnostic NAS benchmark with information of 15,625 neural cell candidates for architecture topology and 32,768 for architecture size on three datasets, and we also provide 13 NAS baselines in a single codebase.
Network Pruning via Transformable Architecture Search
Xuanyi Dong and Yi Yang
in NeurIPS, 2019
[ Paper ] [ Code ] [ Logs ] [ Project ] [ Poster ] [ BibTex ]
Search for the network shape in a differentiable way.
Searching for A Robust Neural Architecture in Four GPU Hours
Xuanyi Dong and Yi Yang
in CVPR, 2019
[ Paper ] [ arXiv ] [ Code ] [ BibTex ]
Propose a gradient-based searching algorithm using differentiable architecture sampling.

Academic Competitions

  • AutoCV2 2019: IMAGE + VIDEO Automated Computer Vision challenge, Rank 4st / 34 teams.
  • TRECVID 2016: Video Localization Task, Rank 1st.
  • ILSVRC 2015: Rank 3rd on Object Detection and 2nd on Object Localization.
  • ACM International Collegiate Programming Contest (ICPC) (2014~2015): 2 Gold and 1 Silver Medals.

Review Services

Journal Reviewer
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPMAI)
International Journal of Computer Vision (IJCV)
IEEE Transactions on Image Processing (TIP)
IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)
Pattern Recognition (PR) [Outstanding Reviewer]
Transactions on Computational Biology and Bioinformatics (TCBB)

Conference Reviewer
Neural Information Processing Systems (NeurIPS) 2019 - 2020
Computer Vision and Pattern Recognition (CVPR) 2019 - 2020
European Conference on Computer Vision (ECCV) 2020
International Conference on Computer Vision (ICCV) 2019
AAAI Conference on Artificial Intelligence (AAAI) 2020 - 2021
ACM Multimedia (ACM MM), 2017 - 2020