Optimizing decision-making in high-stakes settings requires significant advances in mathematical and computational techniques to be applicable to the problems considered and to perform as planned and intended when deployed, being e.g., robust to uncertainty and unmodelled phenomena. In this long-term stream of my work, I have made important strides to advance underlying knowledge and to apply the tools created to make a positive impact through partnerships with Panthera, MGH, the University of Denver, RAND, and the Sitka Sound Science Center. A permeating theme is the need for methods that apply to large scale, sequential decision-making problems, that are often combinatorial in nature.

This research stream is supported in part by my NSF OE grant on conservation, my NSF S&CC grant on landslides, and my US Army Research Laboratory grant on suicide prevention.

My first contribution to methods for robust sequential decision-making is [24], where I propose a data-driven tractable approximation scheme for convex multi-stage robust optimization problems. My approach has attractive scalability properties and is asymptotically optimal. It has important applications in e.g., energy markets [25]. Since then, I have led a research direction dedicated to models and solution approaches for stochastic and robust optimization problems affected by endogenous uncertainty [10, 20, 21, 23]. Notably, I study sequential decision-making problems affected by uncertainty where the time of information discovery is decision-dependent and uncertain parameters only become observable after a costly investment--these are very hard stochastic and robust combinatorial problems and arise in e.g., active preference elicitation [14, 26]. In [23], I devise the first method for modeling and solving stochastic and robust optimization problems with decision-dependent information discovery (DDID) that does not require discretization of the uncertainty space. More recently, in [20], I devise a model and solution approach tailored to robust problems with DDID which substantially outperforms [23] in terms of both interpretability and approximation quality. Recently, in work led by my PhD student Qing Jin, we generalized this method to distributionally robust optimization problems with DDID [10].

In the US, more than 90,000 patients are waiting for transplants and only about 20,000 transplants are performed each year. Annually, nearly 5,000 people die without getting a kidney [5]. This raises hard policy-making questions and very tough decisions for patients. I have taken important steps to assist both groups. In [9], motivated by a collaboration with MGH, I propose a robust optimization formulation and a tractable solution approach for estimating the wait time of a patient who only possesses incomplete information with regard to, e.g., their relative priority, other patients’ preferences, and resource availability, and showcase how my methodology can be applied to assist patients on the waitlist based on data from the United Network for Organ Sharing (UNOS). In [20], I take the point of view of policy-makers who must make hard trade-offs between maximizing efficiency and equity in prioritizing patients for these scarce resources. I demonstrate, based on real data from the UNOS, that my approach can be used to actively elicit the preferences of policy-makers to recommend policies that align with their preferences.

A key strategy to preserve biodiversity consists in purchasing land parcels to obtain a system of habitat reserves. Given limited resources, choosing which parcels to protect poses important challenges. To address this problem, I have partnered with Panthera. In work led by my PhD student Yingxiao Ye [29], we propose a multistage robust optimization formulation with a data-driven uncertainty set which captures the endogenous nature of the binary land use uncertainty and derive a conservative solution approach. Our method reduces conservation loss by 19.46% on average compared to standard approaches.

Suicide is a leading cause of death in the US, causing about one death every 11 minutes [1]. Similarly, landslides are common to almost every state causing from about 25 to 50 deaths each year [4]. An effective strategy to mitigate these risks is to employ social network-based interventions which train a small number of individuals (monitors) in the network to watch out for their piers. In the paper [17], led by my recent PhD graduate Aida Rahmattalabi, we propose a robust optimization model capturing fairness requirements and an approximation scheme for selecting monitors to ensure a maximum number of people are "covered" even when the performance of such monitors is uncertain. We provide a formal analysis of the price of group fairness in this problem and demonstrate the effectiveness of our approach on real-world social networks of homeless youth and of inhabitants of Sitka to mitigate suicide risk and landslide risk, respectively. We are preparing to deploy this intervention in Sitka through our partnership with the Sitka Sound Science Center and in the residence halls at the University of Denver.

1. Facts About Suicide. URL: https://www.cdc.gov/suicide/facts/index.html.

2. Grand Challenges for Social Work. URL: https://grandchallengesforsocialwork.org.

3. LAHSA CES Triage Tool Research & Refinement webpage. URL: https://www.lahsa.org/ documents?id=4370-ces-triage-tool-research-refinement.pdf.

4. Landslides 101. URL: https://www.usgs.gov/programs/landslide-hazards/landslides-101.

5. Too Many Donor Kidneys Are Discarded in U.S. Before Transplantation. URL:https://www.pennmedicine.org/news/news-releases/2020/december/ too-many-donor-kidneys-are-discarded-in-us-before-transplantation.

6. S Aghaei, M J Azizi, and P Vayanos. Learning optimal and fair decision trees for non- discriminative decision-making.In Proceedings of the 33rd AAAI Conference on Artificial Intelligence, 2019.

7. S Aghaei, A Gómez, and P Vayanos. Strong optimal classification trees. Major revision at Operations Research, 2022. URL: https://arxiv.org/abs/2002.09142.

8. M J Azizi, P Vayanos, B Wilder, E Rice, and M Tambe. Designing fair, efficient, and in- terpretable policies for prioritizing homeless youth for housing resources. In Proceedings of the 15th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research, 2018.

9. C Bandi, N Trichakis, and P Vayanos. Robust multiclass queuing theory for wait time esti- mation in resource allocation systems. Management Science, 65(1):152–187, 2018.

10. Q Jin, A Georghiou, P Vayanos, and G Hanasusanto. Distributionally robust optimization with decision-dependentinformation discovery. In preparation for submission to INFORMS Journal on Computing, 2022.

11. N Jo, S Aghaei, J Benson, A Gómez, and P Vayanos. Learning optimal fair classication trees. Under review at second ACM conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO’22), 2022. URL: https://arxiv.org/pdf/2201.09932.pdf.

12. N Jo, S Aghaei, A Gómez, and P Vayanos. Learning optimal prescriptive trees from observa- tional data. Major Revision at Management Science, short version appeared at 2022 AAAI Workshop on AI and Behavior Change, 2021. URL:https://arxiv.org/pdf/2108.13628. pdf.

13. N Jo, B Tang, K Dullerud, S Aghaei, E Rice, and P Vayanos. Evaluating fairness of contextual resource allocationsystems: metrics and impossibility results. In preparation for submission to 37th AAAI Conference on Artificial Intelligence, 2022.

14. C Johnston, S Blessenhohl, and P Vayanos. Preference Elicitation and Aggregation to Aid with Patient Triage during the COVID-19 Pandemic. In preparation for submission to Operations Research; short version appeared in International Conference on Machine Learning (ICML) Workshop on Participatory Approaches to Machine Learning, 2020.

15. N Justin, S Aghaei, A G´omez, and P Vayanos. Optimal robust classification trees. In prepa- ration for submission to Operations Research; short version appeared in AAAI Workshop on Adversarial Machine Learning and Beyond, 2021.

16. A Rahmattalabi, P Vayanos, K Dullerud, and E Rice. Learning resource allocation policies from observational data with an application to homeless services delivery. In 2022 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’22), 2022.

17. A Rahmattalabi, P Vayanos, A Fulginiti, E Rice, B Wilder, A Yadav, and M Tambe. Exploring algorithmic fairness in robust graph covering problems. In Proceedings of the 33rd Conference on Neural Information Processing Systems (NeurIPS), 2019.

18. B Tang, C Koçyigit, and P Vayanos. Learning optimal policies for allocating housing to people experiencinghomelessness from data collected in deployment. In preparation for submission to Management Science, 2022.

19. United-Nations. Resolution adopted by the General Assembly on 6 July 2017, Work of the Statistical Commissionpertaining to the 2030 Agenda for Sustainable Development. 2017.

20. P Vayanos, A Georghiou, and H Yu. Robust optimization with decision-dependent information discovery. Major revisionat Management Science, 2019. URL: https://arxiv.org/pdf/ 2004.08490.pdf.

21. P Vayanos, Q Jin, and G Elissaios. ROC++: Robust Optimization in C++. Forthcoming at INFORMS Journal on Computing, 2022. URL: https://arxiv.org/pdf/2006.08741.pdf.

22. P Vayanos, Q Jin, and G Elissaios. ROCPP Version v2020.0140. INFORMS Journal on Computing, 2022.doi:10.5281/zenodo.6360996.

23. P Vayanos, D Kuhn, and B Rustem. Decision rules for information discovery in multi-stage stochastic programming. InProceedings of the 50th IEEE Conference on Decision and Control, pages 7368–7373, 2011.

24. P Vayanos, D Kuhn, and B Rustem. A constraint sampling approach for multi-stage robust optimization. Automatica, 48(3):459–471, 2012.

25. P Vayanos, W Wiesemann, and D Kuhn. Hedging electricity swing options in incomplete markets. In Proceedings of the 18th IFAC Wold Congress, pages 846–853, 2011.

26. P Vayanos, Y Ye, D McElfresh, J Dickerson, and E Rice. Robust active preference elicitation. Under second round ofreview at Management Science, 2021. URL: https://arxiv.org/pdf/ 2003.01899.pdf.

27. P Vossler, S Aghaei, N Justin, N Jo, A Gómez, and P Vayanos. ODTlearn package v0.1. URL: https://github.com/D3M-Research-Group/odtlearn.

28. P Vossler, S Aghaei, N Justin, N Jo, A Gómez, and P Vayanos. ODTlearn: a Python package for learning optimal decision trees. In preparation for submission to Journal of Machine Learning Research, 2022.

29. Y Ye, C Doehring, A Georghiou, P Vayanos, and H Robinson. Conserving biodiversity via ad- justable robust optimization. In preparation for submission to Management Science; short version appeared in Proc. of the 21st International Conference on Autonomous Agents and Mul- tiagent Systems (AAMAS 2022), Workshop on Autonomous Agents for Social Good (AASG), 2022.