Gauri Jain

research

2026

1. NeurIPS 2026 Workshop

   Spatiotemporal Forecasting of Overdose Calls in Seattle: An Interpretable Framework for Public Health Decision Support

   Ansh Gupta, and Gauri Jain

   NeurIPS Trustworthy AI for Good Workshop, Sep 2026

   Abs Link

   Forecasting where near-term overdose-response demand will occur can help public health agencies direct limited outreach and response resources. We develop an interpretable framework for forecasting next-month Seattle Fire overdose-response calls across 235 census tracts in the observed fire-service footprint in Seattle, Washington. The central model is a four feature Capacity-Aware Response Estimation Explainable Boosting Machine (CARE-EBM), evaluated once on a January–June 2026 rolling-origin holdout comprising 1410 tract-month forecasts. Relative to a three-month average persistence baseline, CARE-EBM reduced RMSE from 0.726 to 0.663 and increased R² from 0.655 to 0.713. Best-Possible Reach (BPR) improved from 0.834 to 0.896 at a capacity of 10 tracts and from 0.793 to 0.858 at 20 tracts. In addition to forecasts, we construct independent K-means typologies of socioeconomic context and historical recorded demand trajectories. Their cross-classification creates an interpretable audit matrix showing how socioeconomic context and recorded demand history intersect across communities. Used alongside CARE-EBM’s additive feature contributions, the matrix gives policymakers a unified view of tract context, forecast behavior, and ranked prioritization. Together, these components provide a practical blueprint for trustworthy public health forecasting by transforming historical response-demand data into capacity-aware, transparent evidence that supports human prioritization of services and reveals how forecast errors and ranked selections are distributed across communities.

2. In Review

   The Prosocial Ranking Challenge: Reducing Polarization on Social Media without Sacrificing Engagement

   Jonathan Stray, Ian Baker, George Beknazar-Yuzbashev, Ceren Budak, Julia Kamin, Kylan Rutherford, Mateusz Stalinski, Tin Acosta, Chris Bail, Michael Bernstein, Mark Brandt, Amy Bruckman, Anshuman Chhabra, Soham De, Kayla Duskin, Sara Fish, Beth Goldberg, Andy Guess, Dylan Hadfield-Menell, Muhammed Haroon, Safwan Hossain, Michael Inzlicht, Gauri Jain, Zaria Jalan, Yanchen Jiang, Alexander P. Landry, Yph Lelkes, Hongfan Lu, Peter Mason, Jennifer McCoy, Smitha Milli, Paul Resnick, Emily Saltz, Martin Saveski, Lisa Schirch, Max Spohn, Siddarth Srinivasan, Alexis Tatore, Luke Thorburn, Joshua A. Tucker, Robb Willer, Magdalena Wojcieszak, Manuel Wüthrich, and Sylvan Zheng

   arXiv preprint, Mar 2026

   Abs Link

   We report the first direct comparisons of multiple alternative social media algorithms on multiple platforms on outcomes of societal interest. We used a browser extension to modify which posts were shown to desktop social media users, randomly assigning 9,386 users to a control group or one of five alternative ranking algorithms which simultaneously altered content across three platforms for six months during the US 2024 presidential election. This reduced our preregistered index of affective polarization by an average of 0.03 standard deviations (p < 0.05), including a 1.5 degree decrease in differences between the 100 point inparty and outparty feeling thermometers. We saw reductions in active use time for Facebook (-0.37 min/day) and Reddit (-0.2 min/day), but an increase of 0.32 min/day (p < 0.01) for X/Twitter. We saw an increase in reports of negative social media experiences but found no effects on well-being, news knowledge, outgroup empathy, perceptions of and support for partisan violence. This implies that bridging content can improve some societal outcomes without necessarily conflicting with the engagement-driven business model of social media.

2025

1. In Review

   CTRL Your Shift: Clustered Transfer Residual Learning for Many Small Datasets

   Gauri Jain, Dominik Rothenhäusler, Kirk Bansak, and Elisabeth Paulson

   arXiv preprint, Aug 2025

   Abs Link

   Machine learning (ML) tasks often utilize large-scale data that is drawn from several distinct sources, such as different locations, treatment arms, or groups. In such settings, practitioners often desire predictions that not only exhibit good overall accuracy, but also remain reliable within each source and preserve the differences that matter across sources. For instance, several asylum and refugee resettlement programs now use ML-based employment predictions to guide where newly arriving families are placed within a host country, which requires generating informative and differentiated predictions for many and often small source locations. However, this task is made challenging by several common characteristics of the data in these settings: the presence of numerous distinct data sources, distributional shifts between them, and substantial variation in sample sizes across sources. This paper introduces Clustered Transfer Residual Learning (CTRL), a meta-learning method that combines the strengths of cross-domain residual learning and adaptive pooling/clustering in order to simultaneously improve overall accuracy and preserve source-level heterogeneity. We provide theoretical results that clarify how our objective navigates the trade-off between data quantity and data quality. We evaluate CTRL alongside other state-of-the-art benchmarks on 5 large-scale datasets. This includes a dataset from the national asylum program in Switzerland, where the algorithmic geographic assignment of asylum seekers is currently being piloted. CTRL consistently outperforms the benchmarks across several key metrics and when using a range of different base learners.

2024

1. PRICAI 2024

   IRL for Restless Multi-armed Bandits with Applications in Maternal and Child Health

   Gauri Jain, Pradeep Varakantham, Haifeng Xu, Aparna Taneja, Prashant Doshi, and Milind Tambe

   PRICAI, Nov 2024

   ⭐ Best Paper Runner-Up ⭐

   Abs Link

   Best Paper Runner-Up

   Public health practitioners often have the goal of monitoring patients and maximizing patients’ time spent in “favorable" or healthy states while being constrained to using limited resources. Restless multi-armed bandits (RMAB) are an effective model to solve this problem as they are helpful to allocate limited resources among many agents under resource constraints, where patients behave differently depending on whether they are intervened on or not. However, RMABs assume the reward function is known. This is unrealistic in many public health settings because patients face unique challenges and it is impossible for a human to know who is most deserving of any intervention at such a large scale. To address this shortcoming, this paper is the first to present the use of inverse reinforcement learning (IRL) to learn desired rewards for RMABs, and we demonstrate improved outcomes in a maternal and child health telehealth program. First we allow public health experts to specify their goals at an aggregate or population level and propose an algorithm to design expert trajectories at scale based on those goals. Second, our algorithm WHIRL uses gradient updates to optimize the objective, allowing for efficient and accurate learning of RMAB rewards. Third, we compare with existing baselines and outperform those in terms of run-time and accuracy. Finally, we evaluate and show the usefulness of WHIRL on thousands on beneficiaries from a real-world maternal and child health setting in India. We publicly release our code here: https://github.com/Gjain234/WHIRL.

2023

1. IJCAI 2023 Workshop

   Inverse Reinforcement Learning for Restless Multi-Armed Bandits with Application to Maternal and Child Health

   Gauri Jain, Pradeep Varakantham, Haifeng Xu, Aparna Taneja, and Milind Tambe

   IJCAI Bridge AI Workshop, Aug 2023

   Abs Link

   We study restless multi-armed bandits (RMABs) in the context of public health, where there is a need to optimize resource allocation decisions. Until now, RMABs typically solve for the optimal planning policy by assuming the reward function in the problem is fully known. However, in this work, we aim to study whether we can learn the most optimal rewards for an RMAB problem given some demonstrated, ideal behavior. To achieve this, we turn to inverse reinforcement learning (IRL) which is a field of study motivated by the desire to understand and learn the underlying reward structure of an agent’s observed behavior. Existing IRL approaches predominantly focus on single agent systems, presenting limitations in dealing with the expansive state spaces characteristic of public health scenarios, where tens of thousands of arms are active simultaneously. We propose a new IRL algorithm specifically for RMAB settings that uses techniques from decision focused learning (DFL) to directly optimize the objective function, allowing for efficient and accurate updates to the learned rewards. We compare our algorithm with the max entropy IRL baseline on runtime and accuracy and find that our algorithm performs better on both metrics. We also propose a framework for how to apply this algorithm in the public health domain where expert trajectories come from domain experts.

2. AAAI 2023 Workshop

   BiomeAzuero2022: A Fine-Grained Dataset and Baselines for Tree Species Classification with Ground Images

   Ziwei Gu^*, Gauri Jain^*, Hongwen Song^*, Isak Diaz Diaz, Margaux Masson-Forsythe, and Jorge Valdes

   AAAI AI for Social Good Workshop, Feb 2023

   Abs Link

   It is becoming increasingly popular for organizations to invest in carbon credits programs involving planting trees to offset their carbon footprint. However, in order to correctly measure carbon sequestration, it is important to know which tree species exist in a given region. This identification task requires very well-trained local botanists who are not only an expensive resource, but also not accurate enough. AIassisted tree classification has become an promising way to solve the problem, but low data quality has been the limiting factor. Our work focuses on determining how to build a high quality dataset of tree parts that can lead to the most accurate tree classification. We contribute BiomeAzuero2022, a publicly available image dataset of 9071 tree images captured in Azuero, Panama with ground truth species labels, organized by different parts of trees. We further provide a methodology for estimating feature importance via machine learning and interpretability methods with the goal of gathering higher-quality image data through a case study on BiomeAzuero2022.

2022

1. Informs 2022

   Sequential Fair Allocation: Achieving the Optimal Envy-Efficiency Tradeoff Curve

   Sean Sinclair, Gauri Jain, Siddhartha Banerjee, and Christina Lee Yu

   Operations Research, Nov 2022

   Abs Link

   We consider the problem of dividing limited resources to individuals arriving over T rounds. Each round has a random number of individuals arrive, and individuals can be characterized by their type (i.e. preferences over the different resources). A standard notion of ’fairness’ in this setting is that an allocation simultaneously satisfy envy-freeness and efficiency. The former is an individual guarantee, requiring that each agent prefers their own allocation over the allocation of any other; in contrast, efficiency is a global property, requiring that the allocations clear the available resources. For divisible resources, when the number of individuals of each type are known upfront, the above desiderata are simultaneously achievable for a large class of utility functions. However, in an online setting when the number of individuals of each type are only revealed round by round, no policy can guarantee these desiderata simultaneously, and hence the best one can do is to try and allocate so as to approximately satisfy the two properties. We show that in the online setting, the two desired properties (envy-freeness and efficiency) are in direct contention, in that any algorithm achieving additive counterfactual envy-freeness up to a factor of LT necessarily suffers a efficiency loss of at least 1/LT. We complement this uncertainty principle with a simple algorithm, HopeGuardrail, which allocates resources based on an adaptive threshold policy and is able to achieve any fairness-efficiency point on this frontier. In simulation results, our algorithm provides allocations close to the optimal fair solution in hindsight, motivating its use in practical applications as the algorithm is able to adapt to any desired fairness efficiency trade-off.

2020

1. NeurIPS 2020

   Adaptive Discretization for Model-Based Reinforcement Learning

   Sean Sinclair, Tianyu Wang, Gauri Jain, Siddhartha Banerjee, and Christina Lee Yu

   Advances in Neural Information Processing Systems 33, May 2020

   Abs Link

   We introduce the technique of adaptive discretization to design an efficient model-based episodic reinforcement learning algorithm in large (potentially continuous) state-action spaces. Our algorithm is based on optimistic one-step value iteration extended to maintain an adaptive discretization of the space. From a theoretical perspective we provide worst-case regret bounds for our algorithm which are competitive compared to the state-of-the-art model-based algorithms. Moreover, our bounds are obtained via a modular proof technique which can potentially extend to incorporate additional structure on the problem. From an implementation standpoint, our algorithm has much lower storage and computational requirements due to maintaining a more efficient partition of the state and action spaces. We illustrate this via experiments on several canonical control problems, which shows that our algorithm empirically performs significantly better than fixed discretization in terms of both faster convergence and lower memory usage. Interestingly, we observe empirically that while fixed discretization model-based algorithms vastly outperform their model-free counterparts, the two achieve comparable performance with adaptive discretization.