---

Research

My broader research focus lies at the intersection of Machine Learning (ML) and Data Mining with the the aim of building practical ML systems that are effective and robust.

[Background] There is a widely observed fact that the data quality sets the ceiling for machine learning. In the real-world, biased samples such as noisy labels are ubiquitous and which would inevitably degenerate the model performance

My research is to develop the effective data mining algorithms, apply the developed techniques to improve the data quality, and further improve the performance of the machine learning algorithms.

---

Publications

(* indicates equal contributions)

• Resolving Training Biases via Influence-based Data Relabeling.
  S. Kong, Y. Shen, L. Huang.
  In Proceedings of 10th International Conference on Learning Representations (ICLR2022 oral，accept rate = 54/3391), [PDF] [Code] [Poster].

• AutoSrh: An Embedding Dimensionality Search Framework for Tabular Data Prediction.
  S. Kong, W. Cheng, Y. Shen, L. Huang.
  IEEE Transactions on Knowledge and Data Engineering (TKDE 2022), [PDF] [Code] [Poster].

• Using All Training Data Effectively for Deep Learning with Noisy Labels.
  S. Kong, Y. Shen, L. Huang.
  Preprint

---

Brief Biography

Shuming Kong is currently a Phd candidate at Shanghai Jiao Tong University, Shanghai, China.

In my free time, I like playing LOL and wathching Premier League. My favorite club is Chelsea and i hope Chelsea could win the Champions League again!

I am always up for new collaborations, drop me an email if you want to chat!