Our group studies distributed learning and decision making, particularly in network environments, and examines how strategic reasoning, information, and incentive design can shape individual and network-level outcomes. We contribute to, and draw from, tools and methods in several disciplines, including network economics, game theory, algorithmic economics, optimization, and control theory. Our work is generously supported by the National Science Foundation (NSF), Amazon, and Cisco.

Some of the current research projects of our group include:

  1. Network Economics: Games on Networks
  2. Economics and Ethics of AI-Driven Decision Making
  3. Reinforcement Learning on Networks
  4. Cyber Security: Incentive Design and Data Analytics

Network Economics: Games on Networks

The outcomes of many social and economic interactions, such as public good provision, spread of research and innovation, and contagion in financial networks, are governed by an underlying network of connections between their users. When participants are rational and self-interested, such interactions can be modeled as network games. Our research in this area is focused on identifying the effects of the network topology on equilibrium analysis and comparative statics of these games. The findings are of importance to network design and policy interventions.

Selected Publications

Provision of Public Goods on Networks: On Existence, Uniqueness, and Centralities. (PDF)
P. Naghizadeh, M. Liu.
In IEEE Transactions on Network Science and Engineering, vol. 5, no. 3, pp. 225-236, 2018.

On the Uniqueness and Stability of Equilibria of Network Games. (PDF)
P. Naghizadeh, M. Liu.
In The 55th Annual Allerton Conference on Communication, Control and Computing, Oct 2017.

Economics and Ethics of AI-Driven Decision Making

The wide-spread adoption of AI and ML tools for decision making raises questions about the fundamental performance, privacy, and fairness guarantees of algorithmic decision making, particularly in the presence of strategic participants. These may include the agents providing the data used in the learning algorithms (e.g., in crowdsourcing), as well as the strategic agents who are governed by such decision rules. Our research in this area analyzes the incentives of such strategic agents and their interactions with AI/ML algorithms, as well as the impacts of data biases on the performance and fairness guarantees of these algorithms.

Selected Publications

Social Bias Meets Data Bias: The Impacts of Labeling and Measurement Errors on Fairness Criteria. (PDF)
Y Liao, P. Naghizadeh.
In The Thirty-Seventh AAAI Conference on Artificial Intelligence (AAAI’23), Feb 2023.

Adaptive Data Debiasing through Bounded Exploration. (PDF)
Y. Yang, Y. Liu, P. Naghizadeh.
In Advances in Neural Informration Processing Systems (NeurIPS’22), Dec 2022.

Fairness Interventions as (Dis)incentives for Strategic Manipulation. (PDF)
X. Zhang, M. Khalili, K. Jin, P. Naghizadeh, M. Liu.
In International Conference on Machine Learning (ICML’22), Jul 2022.

Subsidy Mechanisms for Strategic Classification and Regression Problems. (PDF)
K. Jin, X. Zhang, M. Khalili, P. Naghizadeh, M. Liu.
In ACM conference on Economics and Computation (EC’22), Jul 2022.

Adversarial Contract Design for Private Data Commercialization. (PDF)
P. Naghizadeh, A. Sinha.
In ACM conference on Economics and Computation (EC ‘19), Jun 2019.

Reinforcement Learning on Networks

Reinforcement learning techniques have been receiving increasing attention due to their applications in robotics, games, resource management, and bidding and advertising. Our work in this area focuses on multi-agent reinforcement learning, wherein both agents’ learning and their performance (payoffs) become intertwined. We are particularly interested in the effects of information asymmetry and information sharing in this class of problems. In addition, we have explored the use of reinforcement learning in an edge computing paradigm for enabling secure and personalized augmented reality (AR) experiences.

Selected Publications

Hurts to Be Too Early: Benefits and Drawbacks of Communication in Multi-Agent Learning. (PDF)
P. Naghizadeh, M. Gorlatova, A. Lan, M. Chiang.

Adaptive Fog-Based Output Security for Augmented Reality. (PDF)
S. Ahn, M. Gorlatova, P. Naghizadeh, M. Chiang, P. Mittal.
In ACM SIGCOMM VR/AR Network Workshop, Aug 2018.

Personalized Augmented Reality Via Fog-based Imitation Learning. (PDF)
S. Ahn, M. Gorlatova, P. Naghizadeh, M. Chiang.
In the IEEE Workshop on Fog Computing and the IoT (co-located with IEEE CPS-IoT Week), Apr 2019.

Cyber Security: Incentive Design and Data Analytics

The increasing number and costs of cyber incidents, affecting both industry and government sectors, has drawn increasing attention to the issue of cyber security. Our work in this area identifies incentive mechanisms and policy interventions that can motivate better security decisions by end users and organizations. Our approach combines theoretical models and data analytics to propose such tools for improving cyber security. Specifically, we have explored the design of cyber insurance contracts so as to incentivize improved security investments by organizations, as well as using data analytics for quantifying and predicting cyber security risks.

Selected Publications: Incentive Design

The Impact of Network Design Interventions on CPS Security. (PDF)
P. S. Oruganti, P. Naghizadeh, Q. Ahmed.
In the Control and Decision Conference (CDC’21), Dec 2021.

Using Private and Public Assessments in Security Information Sharing Agreements. (PDF)
P. Naghizadeh, M. Liu.
In IEEE Transactions on Information Forensics and Security, vol. 15, no. 1, pp. 1801-1814, 2020.

Behavioral and Game-Theoretic Security Investments in Interdependent Systems Modeled by Attack Graphs. (PDF)
M. Abdallah, P. Naghizadeh, A. Hota, T. Cason, S. Bagchi, S. Sundaram.
In IEEE Transactions on Control of Network Systems, 7 (4): 1585-1596, 2020.

Designing Cyber Insurance Policies: The Role of Pre-Screening and Security Interdependence. (PDF)
M. Khalili, P. Naghizadeh, M. Liu.
In IEEE Transactions on Information Forensics and Security, 13 (9): 2226-2239, 2018.

Opting out of Incentive Mechanisms: A Study of Security as a Non-Excludable Public Good. (PDF), (online appendix).
P. Naghizadeh, M. Liu.
In IEEE Transactions on Information Forensics and Security, 11 (12): 2790-2803, 2016.

Selected Publications: Data Analytics

Cloudy with a Chance of Breach: Forecasting Cyber Security Incidents. (PDF)
Y. Liu, A. Sarabi, J. Zhang, P. Naghizadeh, M. Karir, M. Bailey, M. Liu.
In the 24th USENIX Security Symposium, Aug 2015.

Risky Business: Fine-grained Data Breach Prediction Using Business Profiles. (PDF)
A. Sarabi, P. Naghizadeh, Y. Liu, M. Liu.
In Journal of Cybersecurity, 2 (1): 15-28, Oxford University Press, 2016.