Zitao Song (宋子韬)

Short Bio

I am a first year Ph.D. student at Department of Computer Science - Purdue University and I am fortunate to be awarded by Lynn Fellowship. I was a Research Associate at the Agent Mediated Intelligence (AMI) Research Group under the supervision of Prof. Bo An at Nanyang Technological University (NTU). I obtained my master's degree in Data Science at The Chinese University of Hong Kong, Shenzhen (CUHK-Shenzhen) under the supervision of Prof. Shuang Li. I obtained my bachelor's degree in Applied Mathematics at Xi'an Jiaotong-Liverpool University (XJTLU) under the supervision of Prof. Jia Meng and Prof. Jionglong Su.

Research

I am currently working with Prof. David Gleich related to efficient post-training with emphasis on matrix reductions and optimizations. I also have broad interests in Sequential Decision Making, Machine Learning, and LLMs. Especially, I focus on the following directions:

• Large-scale optimization for scientific discovery, i.e., biological sequence optimization, material discovery and perference learning.
• RL as Probabilistic Inference, scale and accelerate probabilistic sampling algorithms, i.e., SMC, MCMC, by leveraging world/prior knowledge in LLM/VLMs and diffusion models.

Selected Publications

(* indicates equal contribution, see full list in Google Scholar)

2021 - Present

Latent Logic Tree Extraction for Event Sequence Explanation from LLMs

Zitao Song, Chao Yang, Chaojie Wang, Bo An, Shuang Li.
The Forty-first International Conference on Machine Learning [ICML 2024]

Amortized Network Intervention to Steer the Excitatory Point Processes

Zitao Song, Wendi Ren, Shuang Li.
The Twelfth International Conference on Learning Representations [ICLR 2024]

Attention-based multi-label neural networks for integrated prediction and interpretation of twelve widely occurring RNA modifications

Zitao Song, Daiyun Huang, Bowen Song, Kunqi Chen, Yiyou Song, Gang Liu, Jionglong Su, João Pedro de Magalhães, Daniel J Rigden, Jia Meng.
Nature Communications [Nat. Commun 12(4011), 2021]

Service

Reviewer: ICLR 2024/2025