Yuxin Liu

Ph.D. student, Cornell University

Home

I am a third-year Ph.D. student in Operations Research at the University of Pittsburgh, advised by Prof. M. Amin Rahimian. I hold a B.S. in Economics from Jilin University and an M.E. in Systems Engineering from Tianjin University.

I am a Ph.D. student in Operations Research at the University of Pittsburgh, advised by Prof. M. Amin Rahimian. I develop theory-driven, privacy-aware algorithms for learning and decision-making in social and information networks. My research lies at the intersection of economics and computation, with a focus on sequential social learning, information design, differential privacy, and influence maximization.

A central thread in my work studies how privacy perturbations reshape belief updating, herding behavior, and collective decision efficiency, bridging microeconomic theory with machine learning and algorithmic tools. My empirical and computational projects use large-scale social data, including a 5M-message WhatsApp network, to examine information virality, harmful content diffusion, and behavioral responses in real-world environments.

Recent News

April 2026. Privacy-preserving Information Sharing in Oligopoly Competitions will be presented at the Network Science and Economics (NET ECON) Conference 2026, held at the University of Miami from April 10–12, 2026. Authors: Yuxin Liu and M. Amin Rahimian.
March 2026. I participated in the 2026 IOE-ISyE-MS&E Rising Stars Workshop, held at the University of Michigan on March 27, 2026, where I presented Privacy-preserving Information Sharing in Oligopoly Competitions. Authors: Yuxin Liu and M. Amin Rahimian.
February 2026. Optimal Resolution of a Data Sharing Trilemma: Statistical Power, Sample Complexity, and Privacy Budget was accepted as a poster at the 2026 Privacy and Public Policy Conference (PPPC), held at Georgetown University from February 9–10, 2026.
December 2025. I delivered a tutorial at WINE 2025, The 21st Conference on Web and Internet Economics, held at Rutgers University from December 8–11, 2025, titled Differential Privacy for Strategic Information Sharing and Learning. Tutorial website: Differential Privacy for Strategic Information Sharing and Learning .
October 2025. I chaired the Privacy-Preserving session at the 2025 INFORMS Annual Meeting, held in Atlanta from October 26–29, 2025, and presented Privacy-Aware Sequential Learning.

Publications

Selected Publications:

Structural Dynamics of Harmful Content Dissemination on WhatsApp ICWSM 2026
Yuxin Liu, Kiran Garimella, M. Amin Rahimian.
This paper analyzes large-scale WhatsApp data to study how harmful content, including misinformation, hate speech, and propaganda, spreads in networked communication environments. We combine multimodal features and cascade reconstruction to characterize the structural differences across content types.

Privacy-Aware Sequential Learning FORC 2025
Yuxin Liu, M. Amin Rahimian.
This paper studies sequential social learning when agents use privacy-preserving randomized responses. We show how privacy perturbations reshape belief updating, herding behavior, and information aggregation, and characterize when privacy can improve collective learning efficiency.

Working Papers:

Seeding with Differentially Private Network Information Under review
Yuxin Liu, M. Amin Rahimian, Fang-Yi Yu.
This work develops privacy-preserving algorithms for influence maximization when network information is sensitive or only partially observed. We introduce differentially private seeding mechanisms that balance privacy protection with diffusion performance under constrained feedback.

Differential Privacy for Network Connectedness Indices Under review
T. Rutter, Yuxin Liu, M. Amin Rahimian.
This paper develops differentially private mechanisms for releasing network connectedness indices under edge-adjacent privacy. The goal is to preserve useful measures of group-level connectedness while protecting sensitive relational information in network data.

Privacy-preserving Information Sharing in Oligopoly Competitions Work in progress
Yuxin Liu, M. Amin Rahimian.
This work studies strategic information sharing among competing firms under demand uncertainty and privacy constraints. We characterize how privacy noise and external signals affect firms’ incentives to disclose information and the resulting market outcomes.

Optimal Correlated Noise Design for Privacy–Utility Trade-offs Work in progress
Yuxin Liu, Jie Gao, Ani Sridhar, M. Amin Rahimian.
This work formulates privacy mechanism design as an optimization problem that balances estimation utility and correlation-based leakage in networked data. It develops correlated additive noise mechanisms and convex optimization methods for learning privacy-preserving noise structures under structural constraints.

Modeling User Interaction and Behavioral Dynamics in Large Language Models Work in progress
Yuxin Liu, Kiran Garimella, Yu-Ru Lin, M. Amin Rahimian.
This project builds longitudinal datasets of ChatGPT user interactions to study how prompt complexity, conversation structure, and user behavior evolve over time. It combines LLM-assisted measurement, text representation learning, and statistical modeling to quantify behavioral adaptation in large-scale LLM usage.

Additional Peer-Reviewed Journal Articles:

Research on the mechanism and application of spatial credit risk contagion based on complex network model 2024
J. Ma, Yuxin Liu, L. Zhao, W. Liang.
Managerial and Decision Economics.

Does green credit financing mode with cap-and-trade scheme really benefit all members? 2023
J. Ma, Yuxin Liu, L. Geng.
Managerial and Decision Economics.

Research on optimization of food supply chain considering product traceability recall and safety investment 2022
J. Ma, J. Chen, Yuxin Liu.
Managerial and Decision Economics.

Study on the complexity of channel pricing game in showrooming O2O supply chain 2022
Y. Li, J. Ma, Yuxin Liu.
RAIRO – Operations Research.

Teaching Experience

Teaching Philosophy:

My teaching emphasizes clarity, accessibility, and hands-on problem solving. I aim to help students connect mathematical and statistical concepts with practical applications, especially in data analysis, experimental design, and programming. As a teaching assistant, I focus on creating a supportive learning environment through office hours, detailed feedback, and step-by-step explanations that help students build both conceptual understanding and technical confidence.

Teaching Assistant:

IE 2117: Data for Social Good Fall 2024
University of Pittsburgh, Department of Industrial Engineering
Led office hours and supported undergraduate students with data analysis, statistical modeling, and Python programming assignments.

IE 1072: Design of Experiments and Quality Assurance Fall 2024
University of Pittsburgh, Department of Industrial Engineering
Graded homework, lab assignments, and exams; provided feedback on experimental design, statistical inference, and applied data analysis.

Professional Service

Journal and Conference Reviewer:

Conference Reviewer: NeurIPS, ICLR, AISTATS, WWW.

Journal Reviewer: IEEE Control Systems Letters, ACM Transactions on Knowledge Discovery from Data (TKDD), Omega, IEEE Transactions on Control of Network Systems, IEEE Transactions on Signal and Information Processing over Networks, Future Generation Computer Systems, Physica A, Applied Mathematics and Computation, IEEE Transactions on Network Science and Engineering, Big Data Mining and Analytics.

Conference Organizing Activities:

Session Chair / Organizer, INFORMS Annual Meeting.
Organized and chaired sessions on job market showcase, privacy-preserving methods, and information sharing.

Invited Talks and Presentations:

April 2026. Privacy-preserving Information Sharing in Oligopoly Competitions, Network Science and Economics (NET ECON) Conference 2026, University of Miami, April 10–12, 2026.

March 2026. Privacy-preserving Information Sharing in Oligopoly Competitions, 2026 IOE-ISyE-MS&E Rising Stars Workshop, University of Michigan, March 27, 2026.

February 2026. Optimal Resolution of a Data Sharing Trilemma: Statistical Power, Sample Complexity, and Privacy Budget, poster presentation at the 2026 Privacy and Public Policy Conference (PPPC), Georgetown University, February 9–10, 2026.

December 2025. Differential Privacy for Strategic Information Sharing and Learning, tutorial at WINE 2025, The 21st Conference on Web and Internet Economics, Rutgers University, December 8–11, 2025.
Tutorial website: Differential Privacy for Strategic Information Sharing and Learning .

October 2025. Privacy-Aware Sequential Learning, INFORMS Annual Meeting 2025, Atlanta, October 26–29, 2025.

October 2025. Differentially Private Sequential Learning, NYC Privacy Day 2025, October 17, 2025.

June–July 2025. Differentially Private Sequential Learning, 2025 INFORMS Applied Probability Society Conference, Georgia Tech, Atlanta, June 30–July 3, 2025.
Travel grant recipient.

April 2025. Differentially Private Sequential Learning, 2025 Conference on Network Science and Economics.

March 2025. Differentially Private Sequential Learning and Seeding with Differentially Private Network Information, 6th AAAI Workshop on Privacy-Preserving Artificial Intelligence (PPAI-25), Philadelphia, March 3, 2025.
Travel grant recipient.

August 2024. Structural Dynamics of Harmful Content Dissemination on WhatsApp, 2024 Purdue Operations Conference, Purdue University, August 23–25, 2024.

August 2024. Differentially Private Sequential Learning, Theory and Practice of Differential Privacy (TPDP) 2024 Workshop, Harvard University, August 20–21, 2024.

Membership:

INFORMS, ACM.

Projects

Modeling User Interaction and Behavioral Dynamics in Large Language Models

This project studies how users interact with large language models over time. I build longitudinal datasets of ChatGPT conversations and analyze the temporal evolution of prompt complexity, conversation structure, task type, and user behavior. The project combines LLM-assisted measurement, text representation learning, statistical modeling, and large-scale behavioral analysis to understand adaptation and learning dynamics in LLM usage.

Optimal Correlated Noise Design for Privacy–Utility Trade-offs

This project formulates privacy mechanism design as an optimization problem that balances estimation utility and correlation-based leakage in networked data. I study correlated additive noise mechanisms, including Gaussian and Laplace designs, and develop convex optimization frameworks for learning privacy-preserving noise correlation structures under structural and statistical constraints.

Differential Privacy for Network Statistics and Connectedness Analysis

This project develops differentially private methods for releasing network connectedness indices and related network statistics under edge-level privacy. The work addresses high sensitivity and composition challenges by combining attribute perturbation, debiased estimation, and theoretical guarantees such as consistency and asymptotic normality.

Differentially Private Influence Maximization for Network Interventions

This project studies influence maximization when network information is sensitive, incomplete, or partially observed. I develop privacy-preserving seeding algorithms with theoretical utility–privacy guarantees and evaluate their performance on synthetic and real-world network data, with applications to public health interventions and network-level discovery.

Large-Scale Harmful Content Analysis and Information Diffusion Modeling

This project analyzes large-scale social media data to understand how harmful content spreads in networked communication environments. I design pipelines that integrate text, image, behavioral, and network signals; reconstruct information cascades from partial observations; and quantify diffusion patterns such as breadth, depth, reach, and virality across content types.

Privacy-Aware Sequential Learning and Decision-Making

This project develops theoretical models of sequential social learning under privacy-preserving randomized responses. I study how privacy constraints reshape belief updating, error rates, herding behavior, and information aggregation, and characterize when privacy can improve collective decision efficiency under continuous signals.

CV

Contact

Address:

University of Pittsburgh

Department of Industrial Engineering

3700 O'Hara Street

Pittsburgh, PA 15261

Email: yul435 AT pitt DOT edu