Research interests

This section provides an overview of my research interests, as well some of the relevant publications under each interest.

This does not contain all of my published work; for a comprehensive, chronological list, please go to the publications tab.

Also, here's a link to my Google Scholar profile.

Current research projects (summary)

Generally speaking, my recent research uses tools from control theory, economics, statistics, and optimization to understand the changing role of data in today's interconnected systems. In particular, I'm interested in the following questions:

  • Once machine learning is deployed in real-world applications, how will the distribution of data change? In other words, what is the closed-loop effect of machine learning?

    • The transition from forming predictions to making decisions is the transition from correlation to causation. What structural assumptions are needed to infer causality? How can one efficiently and effectively 'control’ causal structures?

    • When people have a stake in the output of a machine learning algorithm, they will change their behaviors to influence the results. This oftentimes changes the properties of the data distributions that serve as inputs to our machine learning algorithms; how can we incentivize people to behave in socially desirable ways?

  • How can we account for user privacy in these modern systems?

    • Can we come up with methods to isolate the 'useful’ parts of data while projecting away the intrusive parts? In other words, how can we optimize the utility-privacy tradeoff?

    • Can we test proprietary software for privacy properties without direct access to its source code?

    • How can we conceptualize privacy in new application domains such as genetics? What are the correct definitions?

  • How can we model human behavior in a fashion that admits engineering design? What models best capture human behavior, and how can we verify and validate these models experimentally?

Much of the theory I have developed is motivated by applications in cyber-physical systems (CPS) and the Internet of Things (IoT), such as the smart grid, modern transportation networks, or semiautonomous vehicles.

A more detailed breakdown of each of these thrusts, as well as relevant publications, follows below.

Current research projects (detailed)

Learning and strategic behavior

In the process of designing CPS and IoT systems, we must develop an understanding of the effect of data and information structures in settings with strategic agents. In particular, the data that IoT systems collect is oftentimes the result of both aleatoric randomness and the decisions of humans. Using techniques from optimization, information theory, and game theory, we analyze the structure and properties of equilibria when strategic agents interface with machine learning algorithms. Specifically, people can distort, obfuscate, and misrepresent their data for strategic gain as well as out of concerns for privacy. Our analysis identifies the societal effects of this data manipulation, and provides insight into mechanisms which can lessen the deadweight loss.

Relevant publications

  • T. Westenbroek, R. Dong, L. J. Ratliff, S. S. Sastry, “Competitive Statistical Estimation with Strategic Data Sources,” IEEE Transactions on Automatic Control, (early access). [URL]

  • L. J. Ratliff, R. Dong, S. Sekar, T. Fiez, “A Perspective on Incentive Design: Challenges and Opportunities,” Annual Review of Control, Robotics, and Autonomous Systems, vol. 2, 2019. [URL]

  • E. Miehling, R. Dong, C. Langbort, and T. Başar, “The Private Sampling Game,” to appear in the IEEE 58th Conference on Decision and Control, 2019.


When incorporating privacy into the design of CPS, designers must account for the tradeoff between the utility of data and the privacy of users. In our work, we formulate the task of designing privacy-preserving mechanisms as an optimization problem: we design our noise to maximize a privacy metric, subject to a constraint on the operational cost of a dynamical system. Previous work has taken inspiration from channel design problems in information theory to formulate the design of utility-aware, privacy-preserving noise for databases; our work extends this further by incorporating dynamics and the effects of different types of noise on the performance of the closed-loop controller. This general framework for designing privacy-preserving mechanisms can be applied to a wide variety of IoT domains: it can be used to keep user behavior in the home private while still benefiting from advanced metering infrastructures, it can be used to keep user locations private while still providing good traffic estimates, and so on.

Relevant publications

  • R. Dong, A. A. Cárdenas, L. J. Ratliff, H. Ohlsson, S. S. Sastry, “Quantifying the Utility-Privacy Tradeoff in the Internet of Things,” ACM Transactions on Cyber-Physical Systems, vol. 2, issue 2, 2018. [URL]

  • R. Jia, R. Dong, P. Ganesh, S. S. Sastry, C. Spanos, “Towards a Theory of Free-Lunch Privacy in Cyber-Physical Systems,” in the 55th Annual Allerton Conference on Communication, Control, and Computing, 2017. [URL]

  • R. Jia, R. Dong, S. S. Sastry, C. Spanos, “Optimal Sensor-Controller Codesign for Privacy in Dynamical Systems,” in the IEEE 56th Conference on Decision and Control (CDC), 2017. [URL]

  • R. Jia, R. Dong, S. S. Sastry, C. J. Spanos, “Privacy-Enhanced Architecture for Occupancy-based HVAC Control,” in the 8th ACM/IEEE International Conference on Cyber-Physical Systems (ICCPS), 2017. [URL]

  • R. Dong, W. Krichene, A. M. Bayen, and S. S. Sastry, “Differential Privacy of Populations in Routing Games,” in the IEEE 54th Conference on Decision and Control (CDC), 2015. [URL] [Extended Draft]

  • J. Giraldo, A. A. Cárdenas, E. Mojica-Nava, N. Quijano, and R. Dong, “Delay and Sampling Independence of a Consensus Algorithm and its Application to Smart Grid Privacy,” in the IEEE 53nd Annual Conference on Decision and Control (CDC), 2014. [URL]

  • L. J. Ratliff, R. Dong, H. Ohlsson, A. A. Cárdenas, and S. S. Sastry, “Privacy and Customer Segmentation in the Smart Grid,” in the IEEE 53nd Annual Conference on Decision and Control (CDC), 2014. [URL]

  • R. Dong and L. J. Ratliff. “Energy Disaggregation and the Utility-Privacy Tradeoff,” in Big Data Applications in Power Systems, Elsevier, 2017. [URL]

  • R. Dong and L. J. Ratliff. “Privacy in the Internet of Things,” in The Next Wave: The National Security Agency's review of emerging technologies, Vol. 21, No. 2, 2016. [URL]

Human behavior and decision-making testbeds

As IoT systems are gradually incorporated into our day-to-day lives, the nature of people's relationships with their technology will evolve in fashions that are not always a priori predictable. While fields such as economics and discrete choice theory have a set of ‘standard’ assumptions about consumers, these may not necessarily hold for the behaviors of users, or may hold only at certain points in a user's relationship with their technology. One thrust of my research has been focused on designing realistic testbeds to record how people interact with new smart technologies, identify ways in which theoretical models fail to align with the behavior of real users, and provide insight into impactful directions for further research.

Relevant publications

  • K. R. Driggs-Campbell, R. Dong, Ruzena Bajcsy, “Robust, Informative Human-in-the-Loop Predictions via Empirical Reachable Sets,” IEEE Transactions on Intelligent Vehicles, vol. 3, issue 3, 2018. [URL]

  • O. Afolabi, K. R. Driggs-Campbell, R. Dong, M. Kochenderfer, S. S. Sastry, “People as Sensors: Imputing Maps from Human Actions,” in the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018. [URL]

Past research projects

These are some things I have worked on in the past. I'm never opposed to reviving my work in any of them, but my finiteness limits me. If you're a student interested in these areas and joining my group, I'm sure we can work something out.

Energy disaggregation

Energy disaggregation, also known as nonintrusive load monitoring (NILM), is the task of separating aggregate energy data for a whole building into the energy data for individual appliances. Studies have shown that simply providing disaggregated data to the consumer improves energy consumption behavior. Furthermore, this disaggregated data could be used to improve efficiency in energy distribution, provide data to control algorithms in advanced metering infrastructures, or even help target advertising. However, placing individual sensors on every device in a home is not presently a practical solution. I've been working on using a dynamical systems approach to disaggregation, which hopes to improve disaggregation results by modeling the power consumption dynamics of individual devices.

Relevant publications

  • R. Dong, O. Afolabi, L. J. Ratliff, H. Ohlsson, S. S. Sastry, “Energy Disaggregation via Adaptive Filtering,” Under review in the ACM Transactions on Intelligent Systems and Technology.

  • L. J. Ratliff, R. Dong, H. Ohlsson, and S. S. Sastry, “Behavior Modification and Utility Learning via Energy Disaggregation,” in the 19th IFAC World Congress (IFAC WC), 2014. [URL]

  • R. Dong, L. J. Ratliff, H. Ohlsson, and S. S. Sastry, “Fundamental Limits of Nonintrusive Load Monitoring,” in the Conference on High Confidence Networked Systems (HiCoNS), 2014. [URL]

  • R. Dong, L. J. Ratliff, H. Ohlsson, and S. S. Sastry, “Energy Disaggregation via Adaptive Filtering,” in the 51st Annual Allerton Conference on Communication, Control, and Computing, 2013. [URL]

  • R. Dong, L. J. Ratliff, H. Ohlsson, and S. S. Sastry, “A Dynamical Systems Approach to Energy Disaggregation,” in the IEEE 52nd Annual Conference on Decision and Control (CDC), 2013. [URL]

Nonlinear basis pursuit

The field of compressive sensing has generally concerned itself with recovering data from the output of a linear system. Recent work has been done to generalize the methodologies of compressive sensing to more general classes of functions. More details can be found at:

Relevant publications

  • H. Ohlsson, A. Y. Yang, R. Dong, and S. S. Sastry, “Nonlinear Basis Pursuit,” in the 47th Asilomar Conference on Signals, Systems and Computers, 2013. [URL]

  • H. Ohlsson, A. Y. Yang, R. Dong, and S. S. Sastry, “CPRL – An Extension of Compressive Sensing to the Phase Retrieval Problem,” in Neural Information Processing Systems (NIPS), 2012. [URL]

  • H. Ohlsson, A. Y. Yang, R. Dong, and S. S. Sastry, “Compressive Phase Retrieval From Squared Output Measurements Via Semidefinite Programming,” in the 16th IFAC Symposium on System Identification, 2012. [URL]

  • H. Ohlsson, A. Y. Yang, R. Dong, M. Verhaegen, and S. S. Sastry. “Quadratic Basis Pursuit,” in Regularization, Optimization, Kernels, and Support Vector Machines, Chapman & Hall, 2014. [URL]

Ionic polymer-metal composites

During my undergraduate studies, my research centered on system identification of ionic polymer-metal composites. Given the complicated physical dynamics, we attempted to perform black-box modeling, and extract the physical parameters which were most salient to performance. I worked to derive a temperature-dependent model for the IPMC, and verified the model's efficacy in improving results during open-loop control. This work was done under the guidance of Professor Xiaobo Tan.

Relevant publications

  • R. Dong and X. Tan, “Modeling and open-loop control of IPMC actuators under changing ambient temperature,” Smart Materials and Structures, vol. 21, no. 6, 2012. [URL]

  • R. Dong and X. Tan, “Open-loop control of IPMC actuators under varying temperatures,” in Electroactive Polymer Actuators and Devices (EAPAD), 2011. [URL]


These are some cool ideas that tangentially emerged while working on other projects, or early explorations of future research directions.

As mentioned above, this list is not complete; for a more complete list of my publications, please go to the publications tab.

Miscellaneous publications

  • E. Mazumdar, R. Dong, V. Rúbies Royo, C. Tomlin, S. S. Sastry, “A Multi-Armed Bandit Approach for Online Expert Selection in Markov Decision Processes.” [arXiv preprint]

  • R. Dong, E. Mazumdar, S. S. Sastry, “Optimal Causal Imputation for Control.” [arXiv preprint]

  • D. Calderone, R. Dong, S. S. Sastry, “External-Cost Wardrop Equilibria in Routing Games,” in the IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 2017. [URL]

  • D. Scobee, L. J. Ratliff, R. Dong, H. Ohlsson, M. Verhaegen, S. S. Sastry, “Nuclear Norm Minimization for Blind Subspace Identification (N2BSID),” in the IEEE 54th Conference on Decision and Control (CDC), 2015. [URL]

  • S. Krichene, W. Krichene, R. Dong, and A. M. Bayen, “Convergence of heterogeneous distributed learning in stochastic routing games,” in the 53rd Allerton Conference on Communication, Control, and Computing, 2015. [URL]