Jiashuo Liu


Ph.D. Candidate @ Tsinghua CS
Visiting Student Researcher @ Stanford MS&E, Cambridge AI4Med

Image 1 Image 2 Image 2 Image 2 Image 2

liujiashuo77@gmail.com [ CV English ][CV Chinese]


—Fight against the grand and lengthy emptiness with subtle and constant efforts

Bio

Hi, Im Jiashuo! I'm a final-year Ph.D. candidate in the Department of Computer Science & Technology at Tsinghua University, advised by Prof. Peng Cui and Prof. Bo Li. I am cureently visiting Prof. Mihaela van der Schaar's group at Cambridge Centre for AI in Medicine. I was a visiting student researcher at Prof. Jose H. Blanchet's group in the Department of Management Science & Engineering at Stanford University. I work remotely with Prof. Hongseok Namkoong and Prof. Henry Lam at Columbia University.

My research lies at the interface of machine learning, operations research, and health-care applications, including:
(1) Generalization under Distribution Shifts: Tools for understanding real-world distribution shifts and empirical large-scale benchmarks on real-world tabular shifts.
(2) Data Heterogeneity: Algorithms to model and exploit data heterogeneity for better generalization;
(3) Data-Driven Distributionally robust optimization: Addressing the over-pessimism problem in real applications.

Recently, I'm focusing on the robust optimization with/for LLMs, including:
(1) Leverage LLM Embeddings for Robust Tabular Prediction.
(2) Evaluating the Stability of LLMs under Potential Distribution Shifts.

Tutorials/Talks

Stability Evaluation via Distributional Perturbation Analysis

Speaker: Jiashuo Liu

INFORMS'24 Annual Meeting, Seattle, US
Advances in Data-Driven Distributionally Robust Optimization

Data Heterogeneity Analysis for Out-of-Distribution Generalization

Speaker: Peng Cui, Jiashuo Liu

CoLLAs'24 Tutorial: Conference on Lifelong Learning Agents 2024, Pisa, Italy

Model the Data Heterogeneity for Out-of-Distribution Generalization

Speaker: Peng Cui, Jiashuo Liu, Bo Li, Renzhe Xu

SDM'24 Tutorial: SIAM International Conference on Data Mining 2024, Huston, US

Modeling & Exploiting Data Heterogeneity under Distribution Shifts

Speaker: Jiashuo Liu, Tiffany (Tianhui) Cai, Peng Cui, Hongseok Namkoong

NeurIPS'23 Tutorial: Neural Information Processing Systems 2023, New Orleans, US
Selected as Favorite Papers/Presentations (9/3500+) by Two Sigma

Publications

Most recent publications on Google Scholar. * denotes equal contributions.

LLM Embeddings Improve Test-time Adaptation to Tabular Y|X-Shifts

Yibo Zeng*, Jiashuo Liu*, Henry Lam, Hongseok Namkoong

NeurIPS'24 Workshop on Table Representation Learning

Stability Evaluation of Large Language Models via Distributional Perturbation Analysis

Jiashuo Liu, Jiajin Li, Peng Cui, Jose Blanchet

NeurIPS'24 Workshop on Red Teaming GenAI

Bridging Multicalibration and Out-of-distribution Generalization Beyond Covariate Shift

Jiayun Wu, Jiashuo Liu, Peng Cui, Zhiwei Steven Wu

NeurIPS'24: Neural Information Processing Systems 2024

On the Need of a Modeling Language for Distribution Shifts: Illustrations on Tabular Datasets

Jiashuo Liu*, Tianyu Wang*, Peng Cui, Hongseok Namkoong

INFORMS'24 Workshop on Data Science, 2024 (full paper presentation)
NeurIPS'23: Advances in Neural Information Processing Systems, Datasets and Benchmarks Track, 2023
Selected as NeurIPS Favorite Papers/Presentations (9/3500+) by Two Sigma
Under review at Operations Research

Stability Evaluation via Distributional Perturbation Analysis

(α-β order) Jose Blanchet*, Peng Cui*, Jiajin Li*, Jiashuo Liu*

ICML'24: International Conference on Machine Learning 2024

Geometry-Calibrated DRO: Combating Over-Pessimism with Free Energy Implications

Jiashuo Liu, Jiayun Wu, Tianyu Wang, Hao Zou, Bo Li, Peng Cui

ICML'24: International Conference on Machine Learning 2024
Short version at NeurIPS'23, Workshop on Distribution Shifts

Enhancing Distributional Stability among Sub-Populations

Jiashuo Liu, Jiayun Wu, Jie Peng, Xiaoyu Wu, Yang Zheng, Bo Li, Peng Cui

AISTATS'24: International Conference on Artificial Intelligence and Statistics 2024

Towards Robust Out-of-Distribution Generalization Bounds via Sharpness

Yingtian Zou, Kenji Kawaguchi, Yingnan Liu, Jiashuo Liu, Mong-Li Lee, Wynne Hsu

ICLR'24: International Conference on Learning Representations 2024 ((Spotlight))

Measure the Predictive Heterogeneity

Jiashuo Liu*, Jiayun Wu*, Renjie Pi, Renzhe Xu, Xingxuan Zhang, Bo Li, Peng Cui

ICLR'22: International Conference on Learning Representations 2022

Distributionally Robust Learning with Stable Adversarial Training

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li

TKDE'22: IEEE Transactions on Knowledge and Data Engineering 2022

Distributionally Robust Optimization with Data Geometry

Jiashuo Liu*, Jiayun Wu*, Bo Li, Peng Cui

NeurIPS'22: Neural Information Processing Systems 2022 (Spotlight)

Kernelized Heterogeneous Risk Minimization

Jiashuo Liu*, Zheyuan Hu*, Peng Cui, Bo Li, Zheyan Shen

NeurIPS'21: Neural Information Processing Systems 2021

Heterogeneous Risk Minimization

Jiashuo Liu, Zheyuan Hu, Peng Cui, Bo Li, Zheyan Shen

ICML'21: International Conference on Machine Learning 2021

Stable Adversarial Learning under Distributional Shifts

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li, Yishi Lin

AAAI'21: AAAI Conference on Artificial Intelligence

Towards Out-of-Distribution Generalization: A Survey

Jiashuo Liu*, Zheyan Shen*, Yue He, Xingxuan Zhang, Renzhe Xu, Han Yu, Peng Cui

Survey Paper

LLM Embeddings Improve Test-time Adaptation to Tabular Y|X-Shifts

Yibo Zeng*, Jiashuo Liu*, Henry Lam, Hongseok Namkoong

NeurIPS'24 Workshop on Table Representation Learning

Stability Evaluation of Large Language Models via Distributional Perturbation Analysis

Jiashuo Liu, Jiajin Li, Peng Cui, Jose Blanchet

NeurIPS'24 Workshop on Red Teaming GenAI

Bridging Multicalibration and Out-of-distribution Generalization Beyond Covariate Shift

Jiayun Wu, Jiashuo Liu, Peng Cui, Zhiwei Steven Wu

NeurIPS'24: Neural Information Processing Systems 2024

On the Need of a Modeling Language for Distribution Shifts: Illustrations on Tabular Datasets

Jiashuo Liu*, Tianyu Wang*, Peng Cui, Hongseok Namkoong

INFORMS'24 Workshop on Data Science, 2024 (full paper presentation)
NeurIPS'23: Advances in Neural Information Processing Systems, Datasets and Benchmarks Track, 2023
Selected as NeurIPS Favorite Papers/Presentations (9/3500+) by Two Sigma
Under review at Operations Research

Stability Evaluation via Distributional Perturbation Analysis

(α-β order) Jose Blanchet*, Peng Cui*, Jiajin Li*, Jiashuo Liu*

ICML'24: International Conference on Machine Learning 2024

Enhancing Distributional Stability among Sub-Populations

Jiashuo Liu, Jiayun Wu, Jie Peng, Xiaoyu Wu, Yang Zheng, Bo Li, Peng Cui

AISTATS'24: International Conference on Artificial Intelligence and Statistics 2024

Towards Robust Out-of-Distribution Generalization Bounds via Sharpness

Yingtian Zou, Kenji Kawaguchi, Yingnan Liu, Jiashuo Liu, Mong-Li Lee, Wynne Hsu

ICLR'24: International Conference on Learning Representations 2024 ((Spotlight))

Enhancing Distributional Stability among Sub-Populations

Jiashuo Liu, Jiayun Wu, Jie Peng, Xiaoyu Wu, Yang Zheng, Bo Li, Peng Cui

AISTATS'24: International Conference on Artificial Intelligence and Statistics 2024

Measure the Predictive Heterogeneity

Jiashuo Liu*, Jiayun Wu*, Renjie Pi, Renzhe Xu, Xingxuan Zhang, Bo Li, Peng Cui

ICLR'22: International Conference on Learning Representations 2022

Kernelized Heterogeneous Risk Minimization

Jiashuo Liu*, Zheyuan Hu*, Peng Cui, Bo Li, Zheyan Shen

NeurIPS'21: Neural Information Processing Systems 2021

Heterogeneous Risk Minimization

Jiashuo Liu, Zheyuan Hu, Peng Cui, Bo Li, Zheyan Shen

ICML'21: International Conference on Machine Learning 2021

Topology-Aware Dynamic Reweighting for Distribution Shifts on Graph

Weihuang Zheng*, Jiashuo Liu*, Jiaxing Li, Jiayun Wu, Peng Cui, Youyong Kong

Preprint

Geometry-Calibrated DRO: Combating Over-Pessimism with Free Energy Implications

Jiashuo Liu, Jiayun Wu, Tianyu Wang, Hao Zou, Bo Li, Peng Cui

ICML'24: International Conference on Machine Learning 2024
Short version at NeurIPS'23, Workshop on Distribution Shifts

Distributionally Robust Learning with Stable Adversarial Training

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li

TKDE'22: IEEE Transactions on Knowledge and Data Engineering 2022

Distributionally Robust Optimization with Data Geometry

Jiashuo Liu*, Jiayun Wu*, Bo Li, Peng Cui

NeurIPS'22: Neural Information Processing Systems 2022 (Spotlight)

Stable Adversarial Learning under Distributional Shifts

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li, Yishi Lin

AAAI'21: AAAI Conference on Artificial Intelligence

Towards Out-of-Distribution Generalization: A Survey

Jiashuo Liu*, Zheyan Shen*, Yue He, Xingxuan Zhang, Renzhe Xu, Han Yu, Peng Cui

Survey Paper

LLM Embeddings Improve Test-time Adaptation to Tabular Y|X-Shifts

Yibo Zeng*, Jiashuo Liu*, Henry Lam, Hongseok Namkoong

NeurIPS'24 Workshop on Table Representation Learning

Stability Evaluation of Large Language Models via Distributional Perturbation Analysis

Jiashuo Liu, Jiajin Li, Peng Cui, Jose Blanchet

NeurIPS'24 Workshop on Red Teaming GenAI

Bridging Multicalibration and Out-of-distribution Generalization Beyond Covariate Shift

Jiayun Wu, Jiashuo Liu, Peng Cui, Zhiwei Steven Wu

NeurIPS'24: Neural Information Processing Systems 2024

Topology-Aware Dynamic Reweighting for Distribution Shifts on Graph

Weihuang Zheng*, Jiashuo Liu*, Jiaxing Li, Jiayun Wu, Peng Cui, Youyong Kong

Preprint

On the Need of a Modeling Language for Distribution Shifts: Illustrations on Tabular Datasets

Jiashuo Liu*, Tianyu Wang*, Peng Cui, Hongseok Namkoong

INFORMS'24 Workshop on Data Science, 2024 (full paper presentation)
NeurIPS'23: Advances in Neural Information Processing Systems, Datasets and Benchmarks Track, 2023
Selected as NeurIPS Favorite Papers/Presentations (9/3500+) by Two Sigma
Under review at Operations Research

Stability Evaluation via Distributional Perturbation Analysis

(α-β order) Jose Blanchet*, Peng Cui*, Jiajin Li*, Jiashuo Liu*

ICML'24: International Conference on Machine Learning 2024

Geometry-Calibrated DRO: Combating Over-Pessimism with Free Energy Implications

Jiashuo Liu, Jiayun Wu, Tianyu Wang, Hao Zou, Bo Li, Peng Cui

ICML'24: International Conference on Machine Learning 2024
Short version at NeurIPS'23, Workshop on Distribution Shifts

Enhancing Distributional Stability among Sub-Populations

Jiashuo Liu, Jiayun Wu, Jie Peng, Xiaoyu Wu, Yang Zheng, Bo Li, Peng Cui

AISTATS'24: International Conference on Artificial Intelligence and Statistics 2024

Domain-wise Data Acquisition to Improve Performance under Distribution Shift

Yue He, Dongbai Li, Pengfei Tian, Han Yu, Jiashuo Liu, Hao Zou, Peng Cui

ICML'24: International Conference on Machine Learning 2024

Distributionally Generative Augmentation for Fair Facial Attribute Classification

Fengda Zhang, Qianpei He, Kun Kuang, Jiashuo Liu, Long Chen, Chao Wu, Jun Xiao, Hanwang Zhang

CVPR'24: Conference on Computer Vision and Pattern Recognition 2024

Rethinking the Evaluation Protocol of Domain Generalization

Han Yu, Xingxuan Zhang, Renzhe Xu, Jiashuo Liu, Yue He, Peng Cui

CVPR'24: Conference on Computer Vision and Pattern Recognition 2024

Towards Robust Out-of-Distribution Generalization Bounds via Sharpness

Yingtian Zou, Kenji Kawaguchi, Yingnan Liu, Jiashuo Liu, Mong-Li Lee, Wynne Hsu

ICLR'24: International Conference on Learning Representations 2024 ((Spotlight))

Offline Policy Evaluation in Large Action Spaces via Outcome-Oriented Action Grouping

Jie Peng, Hao Zou, Jiashuo Liu, Shaoming Li, Yibao Jiang, Jian Pei, Peng Cui

WWW'23: The ACM Web Conference 2023

Measure the Predictive Heterogeneity

Jiashuo Liu*, Jiayun Wu*, Renjie Pi, Renzhe Xu, Xingxuan Zhang, Bo Li, Peng Cui

ICLR'22: International Conference on Learning Representations 2022

Distributionally Robust Learning with Stable Adversarial Training

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li

TKDE'22: IEEE Transactions on Knowledge and Data Engineering 2022

Distributionally Robust Optimization with Data Geometry

Jiashuo Liu*, Jiayun Wu*, Bo Li, Peng Cui

NeurIPS'22: Neural Information Processing Systems 2022 (Spotlight)

Towards the ultimate PMT waveform analysis for neutrino and dark matter experiments

Dacheng Xu, Benda Xu, Erjin Bao, Yiyang Wu, Aiqiang Zhang, Yuyi Wang, Geliang Zhang, Yu Xu, Ziyi Guo, Jihui Pei, Hanyang Mao, Jiashuo Liu, Zhe Wang, Shaomin Chen

JINST'22: Journal of Instrumentation 2022

Invariant Preference Learning for General Debiasing in Recommendation

Zimu Wang, Yue He, Jiashuo Liu, Wenchao Zou, Philip Yu, Peng Cui

KDD'22: SIGKDD Conference on Knowledge Discovery and Data Mining 2022

Kernelized Heterogeneous Risk Minimization

Jiashuo Liu*, Zheyuan Hu*, Peng Cui, Bo Li, Zheyan Shen

NeurIPS'21: Neural Information Processing Systems 2021

Heterogeneous Risk Minimization

Jiashuo Liu, Zheyuan Hu, Peng Cui, Bo Li, Zheyan Shen

ICML'21: International Conference on Machine Learning 2021

Stable Adversarial Learning under Distributional Shifts

Jiashuo Liu, Zheyan Shen, Peng Cui, Linjun Zhou, Kun Kuang, Bo Li, Yishi Lin

AAAI'21: AAAI Conference on Artificial Intelligence

Triple Generative Adversarial Networks

Chongxuan Li, Kun Xu, Jun Zhu, Jiashuo Liu, Bo Zhang

TPAMI'21: Transactions on Pattern Analysis and Machine Intelligence 2021

Signed Graph Neural Network with Latent Groups

Haoxin Liu, Ziwei Zhang, Peng Cui, Yafeng Zhang, Qiang Cui, Jiashuo Liu, Wenwu Zhu

KDD'21: SIGKDD Conference on Knowledge Discovery and Data Mining 2021

Stable Learning via Differentiated Variable Decorrelation

Zheyan Shen, Peng Cui, Jiashuo Liu, Tong Zhang, Bo Li, Zhitang Chen

KDD'20: SIGKDD Conference on Knowledge Discovery and Data Mining 2020

Achievements

Excellent Comprehensive Scholarship of Tsinghua University , for Ph.D. student, 2022

National Scholarship (Nationwide, Top 1%), for Ph.D. student, 2022

Apple Scholars in AI/ML Nomination (Top-2 in Tsinghua), for Ph.D. student, 2021

Excellent Undergraduate (Top 10% in Tsinghua), 2020

Excellent Academic Scholarship , for undergraduates, 2018, 2019

Excellent Comprehensive Scholarship (Top 5% in Tsinghua), for undergraduates, 2017

Freshman Scholarship (2nd Grade) (Top 0.01% in Nei Mongol), for undergraduates, 2016

Distribution robustness adversarial learning method

Peng Cui, Jiashuo Liu, filed August 30, 2020, issued December 13, 2022.

Certificate

Invariant learning method and device based on heterogeneity hybrid data

Peng Cui, Jiashuo Liu, filed April 28, 2021, and issued January 31, 2023.

Certificate

Services

Journal reviewer: Operations Research, IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), IEEE Transactions on Knowledge and Data Engineering (TKDE), IEEE Transactions on Multimedia (TMM)

Conference reviewer / Program committee: ICLR (2024), NeurIPS (2024, 2023), NeurIPS Datasets & Benchmark (2024), ICML (2022,2023,2024), UAI (2022,2023,2024), AAAI (2022), IJCAI (2022,2023), CVPR (2022,2023, 2024), ECCV (2024), ICCV (2023), CoLLAs (2022,2023,2024), AISTATS (2021,2023,2024), ICDM (2024), NeurIPS DistShift (2023)

Teaching:
Machine Learning Summer Camp for Primary & Middle School Students at Cambridge University (Summer 2024)
Software Engineering at Tsinghua University (Fall 2019, 2020, 2021, 2022, Spring 2022, 2023 TA)
Object-oriented Programming at Tsinghua University (Summer 2022, TA)

Vitæ

Full Resume in PDF.