魏通 Tong Wei Ph.D. candidate LAMDA Group Department of Computer Science & Technology Nanjing University, Nanjing 210023, China. Email: weit [at] lamda.nju.edu.cn

Short Bio

• 2016.09~: Ph.D. candidate, Department of Computer Science & Technology, Nanjing University, China.
  Supervisor: Associate Professor Yu-Feng Li.

• 2012.09~2016.06: Bachelor, Department of Computer Science & Technology, Soochow University, China.

Main Research Interests

• Multi-Label Learning (MLL)

I am currently working on a sub-field of MLL, i.e., eXtreme Multi-label Learning (XML) in which both # of instances and labels are extremely large (# of labels can be more than one million). In this setting, labels are following a long-tailed distribution. Tail labels might be more informative than labels in the head. Subsequently, how to build a classfier to detect tail labels better becomes an emergent challenge.

• Weakly-Supervised Learning (WSL)

Recently, I am intereted in machine learning problems with weakly supervised information, including Semi-Supervised Learning (SSL) and PU learning. Particularly, we are working on improving WSL performance using Deep Neural Networks (DNN).

Resources

• Extreme Classification Repository: for large-scale multi-label datasets and off-the-shelf eXtreme Multi-Label Learning (XML) solvers.


• Mulan Multi-Label Learning Datasets: regular/traditional multi-label learning datasets.


• Workshops: Some workshops regarding XML are listed.


• Related Works: This page categorizes a list of works of my interest, mainly in Multi-Label Learning.


• Call For Papers: This page maintains important dates of recent conferences related to machine learning.


• Hot Topics at Top Conferences: Hot topics at top conferences are extracted and virtualized automatically.


• Softwares of favorite: This page lists some softwares installed on my laptop or phone which I find very useful.

Publications

• Tong Wei, Yu-Feng Li. Does Tail Label Help for Large-Scale Multi-Label Learning. IEEE Transactions on Neural Network and Learning Systems (TNNLS), 2019. [pdf]




• Tong Wei, Lan-Zhe Guo, Yu-Feng Li and Wei Gao. Learning safe multi-label prediction for weakly labeled data. Machine Learning (MLJ), 107(4): 703-725, 2018. [pdf]




• Tong Wei, Wei-Wei Tu, Yu-Feng Li. Learning for Tail Label Data: A Label-Specific Feature Approach. In: Proceedings of the 28th International Joint Conference on Artificial Intelligence (IJCAI'19). Macau, China, 2019. [pdf]




• Tong Wei, Yu-Feng Li. Learning Compact Model for Large-Scale Multi-Label Data. In: Proceedings of the 33rd AAAI Conference on Artificial Intelligence (AAAI'19). Honolulu, Hawaii, 2019. [pdf] [poster] [slides] [code]




• Yu-Feng Li, Hai Wang, Tong Wei, Wei-Wei Tu. Towards Automated Semi-Supervised Learning. In: Proceedings of the 33rd AAAI Conference on Artificial Intelligence (AAAI'19). Honolulu, Hawaii, 2019.[pdf]




• Tong Wei, Yu-Feng Li. Does Tail Label Help for Large-Scale Multi-Label Learning. In: Proceedings of the 27th International Joint Conference on Artificial Intelligence (IJCAI'18). Stockholm, Sweden, 2018. [pdf] [code]

Awards & Honors

• Artificial Intelligence Scholarship, 2018.10

• Huawei Scholarship, 2017.10

• Outstanding Graduate Student, 2016.06

• Excellent Bachelor's Thesis, 2016.05

• ACM-ICPC Asia Regional Contest Silver Medal, 2015.10

Correspondence

Mail

Tong Wei
National Key Laboratory for Novel Software Technology
Nanjing University
Nanjing 210023, China

Laboratory

113, Building of Computer Science and Technology, Xianlin Campus of Nanjing University