Peijie Dong (董佩杰)
My name is Peijie Dong, a Ph.D. student in Data Science and Analysis at the Hong Kong University of Science and Technology (Guangzhou). Under the supervision of Prof. Xiaowen Chu, my research focuses on auto machine learning, deep learning, and model compression. I obtained my B.S. in Software Engineering from Northwest Agriculture & Forestry University supervised by Prof. Hongming Zhang and A.P Yaojun Geng, where I ranked first in my class with a GPA of 3.78/4.0. Additionally, I completed my Master of Engineering at the National University of Defense Technology under the guidance of Advisor A.P. Xin Niu and Zhiliang Tian. Apart from my primary research interests, I also have a keen interest in large language models and diffusion models.
Feel free to add me on WeChat with the ID “topeijie” for further discussions and conversations. I look forward to connecting with you!
Research Interests
- Automated Machine Learning (AutoML)
- Model Compression including Knowledge Distillation, Quantization, Pruning and NAS.
- Large Language Model (LLM)
Publications
- Peijie Dong, Lujun Li, Zimian Wei. DisWOT: Student Architecture Search for Distillation without Training, The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.
- Peijie Dong, Lujun Li, Zimian Wei, Xin Niu$^*$, Zhiliang Tian, Hengyue Pan. EMQ: Evolving Training-free Proxies for Automated Mixed Precision Quantization, International Conference on Computer Vision (ICCV), 2023
- Peijie Dong, Xin Niu, Zhiliang Tian, etc. Progressive Meta-Pooling Learning for Lightweight Image Classification Model, IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023.
- Peijie Dong, Xin Niu, Lujun Li, etc. RD-NAS: Enhancing One-shot Supernet Ranking Ability via Ranking Distillation, IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023.
- Peijie Dong, Xin Niu, Hengyue Pan, etc. AutoRF: Auto Learning Receptive Fields with Spatial Pooling, 29th International Conference on Multimedia Modeling (MMM), 2022.
- Peijie Dong, Xin Niu$^*$, Lujun Li, etc. Prior-Guided One-shot Neural Architecture Search, Computer Vision and Pattern Recognition Third workshop on Neural Architecture Search (CVPRW), 2022.
- Lujun Li, Peijie Dong, Zimian Wei, Yang Ya. Automated Knowledge Distillation via Monte Carlo Tree Search, International Conference on Computer Vision (ICCV), 2023
Competitions & Awards
- 3rd place 2022 Computer Vision and Pattern Recognition (CVPR) Third Workshop on NAS
- 1st place 2022 BDCI Retail Product Recognition based on MindSpore (CCF Big Data & Computing Interlligence Context 2022)
- 1st place 2022 DCIC Intelligent Ship Detection Competition (Digital China Innovation Contest 2022)
- 2st place 2022 DCIC Intelligent Cattle Segmentation Competition (Digital China Innovation Contest 2022)
- 1st place 2022 Baidu AI Competition - Blurred Document Image Recovery
Experences
- Refactored the design of NAS part (Mutable and Mutator, etc).
- Gained proficiency in Python development tools such as Git, typehint, CI, etc.
- Reproduce the one-shot transformer search with weight entanglement.
- Refactor the design of Neural Architecture Search.
- Internship in OpenMMLab MMRazor.
- Modularized Neural Architecture Search framework.
- Support 5 NAS benchmarks, 6 NAS algorithms.
- Encompass RD-NAS and Meta-Pooling
- The 3rd solution of the Neural Architecture Search Workshop.
- The 1st solution of BDCI Retail Product Recognition based on MindSpore
Academic Services
Reviewer:
- IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023 and 2024
- Conference on Neural Information Processing Systems (NeurIPS), 2023
- IEEE Computational Intelligence Magazine (CIM), 2023
- International Conference on Learning Representations (ICLR), 2024
- Neural Information Processing Systems (NeurIPS), 2023
- IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024
- IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2024
- Information Fusion, 2024
- Neural Network, 2024
- The European Conference on Computer Vision (ECCV), 2024