Jialin Liu

My research interest lies in the intersection of applied mathematics and machine learning (ML). For example, I'm currently working on “Learning to Optimize (L2O).” In contrast to traditional optimization algorithms based on worst-case analysis, L2O approaches using ML techniques to discover novel schemes from datasets comprised of optimization instances. These ML-driven strategies have the potential to detect “shortcuts” that outperform the efficiency of conventional algorithms. Understanding these learned algorithms and ensuring their performance on instances absent from the training set are critical challenges in this topic. My research is committed to establishing the theoretical foundations for such approaches, with the aim to enhance the interpretability and generalization capabilities of ML-based optimization algorithms, while preserving the benefits of L2O.

Some selected papers:

(With X. Chen, Z. Wang, W. Yin, and H. Cai). “Towards Constituting Mathematical Structures for Learning to Optimize.” International Conference on Machine Learning (ICML), 2023. (pdf)
(With Z. Chen, X. Wang, J. Lu, and W. Yin). “On Representing Linear Programs by Graph Neural Networks.” International Conference on Learning Representations (ICLR), 2023. (pdf, software, Spotlight paper)
(With Z. Chen, X. Wang, J. Lu, and W. Yin). “On Representing Mixed-Integer Linear Programs by Graph Neural Networks.” International Conference on Learning Representations (ICLR), 2023. (pdf, software)
(With X. Chen, Z. Wang, and W. Yin). “ALISTA: Analytic Weights Are As Good As Learned Weights in LISTA.” International Conference on Learning Representations (ICLR), 2019. (pdf, poster, software, Top-rated paper)

Jialin Liu

Educational Experiences

Awards & Honors

Research Interests and Selected Papers