Profile Picture

Jackson A Killian

  jkillian+at+g+dot+harvard+dot+edu

  [CV]

  [Google Scholar]

  @JacksonAKillian

  [killian-34]


Harvard Logo

News

Feb '23: Presented at AAAI about ongoing collaboration with ARMMAN, a maternal telehealth nonprofit in India. Technical focus on scaling robust planning for restless bandits up to 100,000s of arms. I also chaired a session during the inspiring Innovative Applications of AI (IAAI) conference!


Nov '22: Continued Research with Verily! Continuing to develop algorithms for targeted health interventions to support chronic disease management. Work is under submission.


June '22: Research at Google! Working as a student researcher with Google's AI for Social Good team, designing algorithms for targeted health interventions to support chronic disease management.


June '22: Accepted to UAI'22! We designed the first algorithms for robust planning in RMABs, powered by our new reinforcement learning algorithm for RMABs and a double oracle framework.


June '21: Accepted to KDD'21! Introduces algos for learning multi-action RMABs online. Alg. 1 (MAIQL) finds an index policy with convergence guarantees. Alg. 2 (LPQL) directly solves the Lagrangian relaxation, and is built for speed+robustness. A step toward RMABs for the real-world, where dynamics are often unknown.


Dec '20: Accepted to AAMAS'21! Work extending Restless Bandits past the paradigm of "to act or not to act" by allowing for multiple action choices per arm. Relevant for planning in domains where multiple intervention options are available to one resource-constrained planner (e.g., community health care). [publication] [code]


Sept '20: Accepted to NeurIPS! Co-first author work with Aditya Mate developing theory for Collapsing Bandits, a new subclass of Restless Bandits. Simulation evaluations on a real-world-derived public health challenge. Extensions to follow [publication] [code]


Sept '20: Accepted to PNAS! Our COVID-19 modeling work of between-population variation of disease dynamics by accounting for country-specific demographics, contact patterns, comorbidities, and household structures [publication] [code]


April '20: Live interview on ABC-7 WJLA about our work modeling COVID-19 [video]


April '20: (COVID-19) See our collaborative work with The Daily Beast to evaluate Georgia's proposed reopening policies. Great experience working the journalists on The Daily Beast team! [article] [preprint]

Jackson presenting at workshop in India

I am a PhD student in Computer Science at Harvard University. Advised by Prof. Milind Tambe, I study how AI techniques can solve challenges in public health, with a technical focus in machine learning and sequential decision making. Using these approaches, I have developed predictive models and sequential planning tools for community health workers in India combatting tuberculosis. Part of this work I conducted from Bangalore while on a research internship with Microsoft Research India, where I was grateful to be advised by Dr. Amit Sharma. In my graduate work, I am proud and fortunate to be supported by a 2019-'24 National Science Foundation Graduate Research Fellowship.


In the summer of 2022 I worked as a student researcher with Google's AI for Social Good organization, developing models of disease progression as well as algorithms for planning targeted health interventions to support a diabetes management platform. I was grateful to be advised by Philip Nelson and Manish Jain. I am now continuing this work as an intern with Verily's Data Science team, advised by Yugang Jia.


Previously, I have also worked on building agent-based models for simulating non-pharmacological interventions for preventing the spread of COVID-19 and designed machine learning models for detecting sobriety using mobile health data (check out the dataset!) During my undergraduate degree at The Ohio State Univeristy, I studied biophysics as a Pelotonia fellow, where I am grateful to have been advised by Dr. Ralf Bundschuh and Dr. Pearlly Yan.


Outside of research, I have worked to build a stronger community around AI for Social Good work. I previously co-directed the Harvard AI in Healthcare Group, a cross-sectional organization that brings together students, researchers and entrepreneurs from across disciplines to explore opportunities and hear from leading experts in this exciting space. I also co-organized the Harvard CRCS Rising Stars Workshops in 2020 and 2021 which highlighted and provided networking and mentorship opportunities to budding junior researchers dedicated to studying AI for Social Impact.


Publications

* equal contribution
  • IJCAI 2023: Gordon L, Behari N, Collier S, Bondi-Kelly E, Killian JA, Ressijac C, Boucher P, Davies A, Tambe M. Find Rhinos without Finding Rhinos: Active Learning with Multimodal Imagery of South African Rhino Habitats. International Joint Conference on Artificial Intelligence. 2023 August. [preprint]
  • IJCAI 2023: Danassis P, Verma S, Killian JA, Taneja A, Tambe M. Limited Resource Allocation in a Non-Markovian World: The Case of Maternal and Child Healthcare. International Joint Conference on Artificial Intelligence. 2023 August. [preprint]
  • AAMAS 2023: Biswas A, Killian JA, Rodriguez Diaz P, Ghosh S, Tambe M. Fairness for Workers Who Pull the Arms: An Index Based Policy for Allocation of Restless Bandit Tasks. Conference on Autonomous Agents and Multiagent Systems. 2023 May. [preprint]
  • AAAI 2023: Killian JA,* Biswas A,* Xu L,* Verma S,* Nair V, Taneja A, Hegde A, Madhiwalla N, Rodriguez Diaz P, Johnson-Yu S, Tambe M. Robust Planning over Restless Groups: Engagement Interventions for a Large-Scale Maternal Telehealth Program. AAAI Conference on Artificial Intelligence. 2023 Feb. [publication]
  • AAAI 2023: Rodriguez Diaz P, Killian JA, Xu L, Taneja A, Suggala AS, Tambe M. Flexible Budgets in Restless Bandits: A Primal-Dual Algorithm for Efficient Budget Allocation. AAAI Conference on Artificial Intelligence. 2023 Feb. [publication]
  • UAI 2022: Killian JA, Xu L, Biswas A, Tambe M. Restless and Uncertain: Robust Policies for Restless Bandits via Deep Multi-Agent Reinforcement Learning. Conference on Uncertainty in Artificial Intelligence. 2022 August. [publication]
  • AAMAS 2022: Ou HC, Siebenbrunner C, Killian JA, Brooks MB, Kempe D, Vorobeychik Y, Tambe M. Networked Restless Multi-Armed Bandits for Mobile Interventions. Conference on Autonomous Agents and Multiagent Systems. 2022 May. [publication]
  • KDD 2021: Killian JA, Biswas A, Shah S, Tambe M. Q-Learning Lagrange Policies for Multi-Action Restless Bandits. Conference on Knowledge Discovery & Data Mining. 2021 August. [publication] [code]
  • AIES 2021: Bondi E,* Xu L,* Acosta-Navas D, Killian JA. Envisioning Communities: A Participatory Approach Towards AI for Social Good. AAAI/ACM Conference on AI, Ethics, and Society. 2021 July. [publication]
  • AAMAS 2021: Killian JA, Perrault A, Tambe M. Beyond "To Act or Not to Act": Fast Lagrangian Approaches to General Multi-Action Restless Bandits. Conference on Autonomous Agents and Multiagent Systems. 2021 May. [publication] [code]
  • NeurIPS 2020: Mate A,* Killian JA,* Xu H, Perrault A, Tambe M. Collapsing Bandits and Their Application to Public Health Interventions. Neural Information Processing Systems (NeurIPS). 2020 December. [publication] [code]
  • PNAS 2020: Wilder B, Charpignon M, Killian JA, Ou HC, Mate A, Jabbari S, Perrault A, Desai A, Tambe M, Majumder MS. Modeling between-population variation in COVID-19 dynamics in Hubei, Lombardy, and New York City. Proceedings of the National Academy of Sciences. 2020 September 24. [publication]
  • SSRN 2020: Killian JA, Charpignon M, Wilder B, Perrault A, Tambe M, Majumder MS. Evaluating COVID-19 Lockdown and Reopening Scenarios For Georgia, Florida, and Mississippi. Available at SSRN 3598744. 2020 May 12. [workshop paper]
  • SSRN 2020: Mate A, Killian JA, Wilder B, Charpignon M, Awasthi A, Tambe M, Majumder MS. Evaluating COVID-19 Lockdown Policies For India: A Preliminary Modeling Assessment for Individual States. Available at SSRN 3575207. 2020. [preprint]
  • KDD 2019: Killian JA, Wilder B, Sharma A, Choudhary V, Dilkina B, Tambe M. Learning to Prescribe Interventions for Tuberculosis Patients using Digital Adherence Data. ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2019. doi: 10.1145/3292500.3330777 [publication]
  • IJCAI-KDH 2019: Killian JA, Passino K, Nandi A, Madden D, Clapp J. Learning to Detect Heavy Drinking Episodes Using Smartphone Accelerometer Data. Proceedings of the 4th International Workshop on Knowledge Discovery in Healthcare Data. 2019. http://ceur-ws.org/Vol-2429/paper6.pdf [publication v1] [publication v2] [dataset]
  • BMC Genomics 2018: Killian JA, Topiwala T, Pelletier A, Frankhouser D, Yan P, Bundschuh R. FuSpot: A Web-based Tool for Visual Evaluation of Fusion Candidates. BMC Genomics. 2018. 19:139. doi: 10.1186/s12864-018-4486-3 [publication] [slides] [website]
  • Thyroid 2018: He H, Li W, Yan P, Bundschuh R, Killian JA, Labanowska J, Brock P, et al. Identification of a Recurrent LMO7-BRAF Fusion in Papillary Thyroid Carcinoma. Thyroid. 2018. doi: 10.1089/thy.2017.0258. [publication]

Media

  • Aria Bendix. "Four Days of Work, Followed by 10 Days of Lockdown Could Help Prevent Another Wave of Infections." Business Insider France, May 25, 2020. [article]
  • Leah Burrows. "What is the Right Strategy to Limit the Spread of COVID-19?" Medical Xpress, May 4, 2020. [article]
  • "Models for the Spread of COVID-19." Live interview on ABC-7 WJLA. April 30, 2020. [video]
  • Amanda Mull. "Georgia’s Experiment in Human Sacrifice." The Atlantic, April 29, 2020. [article]
  • William Bredderman and Olivia Messer. "New Model Shows How Deadly Lifting Georgia’s Lockdown May Be." Daily Beast, April 28, 2020. [article]
  • Subhra Priyadarshini. "Model Finds 'Middle Ground' for India's Lockdown Exit." Nature India. April 27, 2020. [article]
  • Pelotonia. "Pelotonia Investment Report (2017)." The James Ohio State University Comprehensive Cancer Center. May 2017. [report]