Arpit Agarwal Postdoctoral Researcher FAIR Labs, Meta Email: agarpit [at] outlook.com Link to: CV (last updated: Jan 2023)

About Me

I am currently a postdoctoral researcher at FAIR Labs (Meta) working with Max Nickel on socially responsible recommendation systems. I was previously a postdoctoral fellow at the Data Science Institute at Columbia University hosted by Prof. Yash Kanoria and Prof. Tim Roughgarden. My research interests lie at the intersection of Machine Learning, Economics and Computation, and Theoretical Computer Science. Specifically, I am interested in the human decision-making process and its interaction with machine learning algorithms-- how humans choose items based on recommendations, how can we learn from their behaviour and recommend better items, and how do these recommendations influence their tastes in the long-term?

Prior to joining Columbia, I completed my PhD from the Department of Computer & Information Science at University of Pennsylvania, under the guidance of Prof. Shivani Agarwal. Before that I completed my masters in computer science and engineering at Indian Institute of Science. I also spent a semester visiting Prof. David Parkes at Harvard, working on the problem of multi-task peer prediction.

Updates

• I am co-organizing an ICML'24 workshop titled "Humans, Algorithmic Decision-Making and Society: Modeling Interactions and Impact". More details to follow soon.
  


• I will be joining the CSE department at IIT Bombay as an Assistant Professor in Fall 2024.
  

Research Publications

• System-2 Recommenders: Disentangling Utility and Engagement in Recommendation Systems via Temporal Point-Processes
  Arpit Agarwal, Nicolas Usunier, Alessandro Lazaric, Maximilian Nickel.
  To Appear at ACM Conference on Fairness, Accountability, and Transparency (FAccT) 2024.


• Misalignment, Learning, and Ranking: Harnessing Users Limited Attention
  Arpit Agarwal, Rad Niazadeh, Prathamesh Patil (alphabetical order) .
  [arXiv preprint]


• Semi-Bandit Learning for Monotone Stochastic Optimization
  Arpit Agarwal, Rohan Ghuge, Viswanath Nagarajan.
  [arXiv preprint]


• Learning-Augmented Dynamic Submodular Maximization
  Arpit Agarwal, Eric Balkanski.
  [arXiv preprint]


• Online Recommendations for Agents with Discounted Adaptive Preferences
  Arpit Agarwal, William Brown (alphabetical order) .
  ALT 2024. [paper]


• Parallel Approximate Maximum Flows in Near-Linear Work and Polylogarithmic Depth
  Arpit Agarwal, Sanjeev Khanna, Huan Li, Prathamesh Patil, Chen Wang, Nathan White, Peilin Zhong (alphabetical order) .
  SODA 2024. [paper]


• When Can We Track Significant Preference Shifts in Dueling Bandits?
  Joe Suk, Arpit Agarwal.
  NeurIPS 2023. [paper]


• Diversified Recommendations for Agents with Adaptive Preferences
  Arpit Agarwal, William Brown (alphabetical order) .
  NeurIPS 2022. [paper]


• Sublinear Algorithms for Hierarchical Clustering
  Arpit Agarwal, Sanjeev Khanna, Huan Li, Prathamesh Patil (alphabetical order) .
  NeurIPS 2022. [paper]


• An Asymptotically Optimal Batched Algorithm for the Dueling Bandit Problem
  Arpit Agarwal, Rohan Ghuge, Viswanath Nagarajan (alphabetical order) .
  NeurIPS 2022. [paper]


• A Sharp Memory-Regret Trade-Off for Multi-Pass Streaming Bandits
  Arpit Agarwal, Sanjeev Khanna, Prathamesh Patil (alphabetical order) .
  COLT 2022. [paper]


• Batched Dueling Bandits
  Arpit Agarwal, Rohan Ghuge, Viswanath Nagarajan (alphabetical order) .
  ICML 2022. Long presentation (top 2% of submissions). [arXiv preprint].


• PAC Top-$k$ Identification under SST in Limited Rounds
  Arpit Agarwal, Sanjeev Khanna, Prathamesh Patil (alphabetical order) .
  AISTATS 2022. [paper]


• Stochastic Dueling Bandits with Adversarial Corruption
  Arpit Agarwal, Shivani Agarwal, Prathamesh Patil (alphabetical order) .
  ALT 2021. [paper]


• Choice Bandits
  Arpit Agarwal, Nicholas Johnson, Shivani Agarwal.
  NeurIPS 2020. [paper] [supplemental]


• Rank Aggregation from Pairwise Comparisons in the Presence of Adversarial Corruptions
  Arpit Agarwal, Shivani Agarwal, Sanjeev Khanna, and Prathamesh Patil (alphabetical order) .
  ICML 2020. [paper]


• Peer Prediction with Heterogeneous Users.
  Arpit Agarwal, Debmalya Mandal, David C. Parkes , and Nisarg Shah (alphabetical order) .
  ACM Transactions on Economics and Computation (TEAC) 2020. [paper]
  Supercedes the EC-17 paper below.


• Stochastic Submodular Cover with Limited Adaptivity.
  Arpit Agarwal, Sepehr Assadi, and Sanjeev Khanna (alphabetical order) .
  SODA 2019. [paper] [arXiv version]
  


• Accelerated Spectral Ranking.
  Arpit Agarwal, Prathamesh Patil, and Shivani Agarwal.
  ICML 2018. [paper]
  


• Learning with Limited Rounds of Adaptivity: Coin Tossing, Multi-Armed Bandits, and Ranking from Pairwise Comparisons.
  Arpit Agarwal, Shivani Agarwal, Sepehr Assadi, and Sanjeev Khanna (alphabetical order) .
  COLT 2017. [paper]
  


• Peer Prediction with Heterogeneous Users.
  Arpit Agarwal, Debmalya Mandal, David C. Parkes , and Nisarg Shah (alphabetical order) .
  EC 2017. [paper]
  


• Informed Truthfulness in Multi-Task Peer Prediction.
  Victor Shnayder, Arpit Agarwal, Rafael Frongillo, and David C. Parkes .
  EC 2016. [paper] [arXiv version]
  A short version appeared in HCOMP Workshop on Mathematical Foundations of Human Computation, 2016


• On Consistent Surrogate Risk Minimization and Property Elicitation.
  Arpit Agarwal and Shivani Agarwal.
  COLT 2015. [paper]
  
  
• GEV-Canonical Regression for Accurate Binary Class Probability Estimation when One Class is Rare.
  Arpit Agarwal, Harikrishna Narasimhan, Shivaram Kalyanakrishnan and Shivani Agarwal.
  ICML 2014. [paper]