Xujie Si
About
Appointed Canada CIFAR AI Chair – 2021
Xujie Si is a Canada CIFAR AI Chair at Mila, and an assistant professor in the Department of Computer Science at the University of Toronto.
Si’s research lies at the intersection of programming languages and artificial intelligence. He is interested in developing AI-based techniques to democratize programming for non-professional programmers, and to help professional programmers build secure and reliable software with less effort. He has developed intelligent systems that automatically synthesize programs from input and output examples, mine correct/incorrect software usages from massive codebases, efficiently diagnose software vulnerabilities through a small amount of user feedback, and formally verify software correctness without requiring human annotations.
He is also interested in leveraging formal methods and programming principles to improve AI techniques, especially in terms of interpretability, robustness, and data efficiency.
Awards
- ACM SIGPLAN Distinguished Paper Award, PLDI, 2019
- CSAW Best Applied Security Paper finalist, 2016
Relevant Publications
- Geng, C., Le, N., Xu, X., Wang, Z., Gurfinkel, A., & Si, X. (2023). Towards Reliable Neural Specifications. International Conference on Machine Learning (pp. 11196-11212).
- Li, Z., Guo, J., Jiang, Y., & Si, X. (2023). Learning Reliable Logical Rules with SATNet. Advances in Neural Information Processing Systems, 36.
- Huang, J., Li, Z., Chen, B., Samel, K., Naik, M., Song, L., & Si, X. (2021). Scallop: From probabilistic deductive databases to scalable differentiable reasoning. In Advances in Neural Information Processing Systems 34 (pp. 25134-25145).
Si, X., Raghothaman, M., Heo, K., & Naik, M. (2019). Synthesizing datalog programs using numerical relaxation. arXiv preprint arXiv:1906.00163.
Heo, K., Raghothaman, M., Si, X., & Naik, M. (2019). Continuously reasoning about programs using differential bayesian inference. In Proceedings of the 40th ACM SIGPLAN Conference on Programming Language Design and Implementation (pp. 561-575).