Tejas Pagare

About

Advisors

Maria-Florina Balcan & Karan Singh

Interests

AI + Economics · Multi-Agent AI Systems

Previously

AI Researcher, Honda R&D Tokyo

Education

Electrical Engineering, IIT Bombay

Publications

1. Learning to Price with PersuasionMaria-Florina Balcan, Tejas Pagare, Karan Singh. Conference on Neural Information Processing Systems (NeurIPS 2026)Spotlight(top 3.7% of accepted papers)
   [arXiv][pdf][abstract][bibtex]

   Motivated by modern marketplaces, where the platform or the seller routinely gathers detailed user profiles, we study a novel learning theoretic model that simultaneously involves information and mechanism design. Specifically, we consider the economic setting recently introduced by Bergemann et al. (2022), where in addition to the menu of quality-price pairs, the seller offers information on the value of the match between product quality and buyer's taste via a signaling scheme. We relax the assumption that the seller knows the buyers' belief about the distribution of tastes and study the sample requirements of designing a revenue maximizing scheme. We consider both the batch setting where we have access to data from a set of i.i.d. buyers and an online demand query model where we observe the buyers' behaviors to seller's schemes. Despite the apparent non-convexity of the problem, we also give the first FPTAS to compute a scheme that maximizes the revenue within an arbitrarily small additive loss, which was left open by Bergemann et al. (2022). Overall, this brings a new learning perspective in asymmetric economic settings where buyers and sellers know different types of information.

   copy@inproceedings{balcan2026learning,
     title     = {Learning to Price with Persuasion},
     author    = {Balcan, Maria-Florina and Pagare, Tejas and Singh, Karan},
     booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
     year      = {2026}
   }

2. Data-Driven Estimation of Tail Probabilities under Varying Distributional AssumptionsDohyun Ahn, Sandeep Juneja, Tejas Pagare, Shreyas Samudra. Winter Simulation Conference (WSC 2025)Invited
   [IEEE][bibtex]

   copy@inproceedings{ahn2025datadriven,
     title     = {Data-Driven Estimation of Tail Probabilities under Varying Distributional Assumptions},
     author    = {Ahn, Dohyun and Juneja, Sandeep and Pagare, Tejas and Samudra, Shreyas},
     booktitle = {Proceedings of the Winter Simulation Conference (WSC)},
     year      = {2025}
   }

3. Optimal Algorithms for Bandit Learning in Matching MarketsTejas Pagare, Agniv Bandyopadhyay, Sandeep Juneja. NeurIPS Workshop on Dynamics at the Frontiers of Optimization, Sampling, and Games (2025)
   [OpenReview][arXiv][pdf][slides][abstract][bibtex]

   We study the problem of pure exploration in matching markets under uncertain preferences, where the goal is to identify a stable matching with confidence parameter $δ$ and minimal sample complexity. Agents learn preferences via stochastic rewards, with expected values indicating preferences. This finds use in labor market platforms like Upwork, where firms and freelancers must be matched quickly despite noisy observations and no prior knowledge, in a stable manner that prevents dissatisfaction. We consider markets with unique stable matching and establish information-theoretic lower bounds on sample complexity for (1) one-sided learning, where one side of the market knows its true preferences, and (2) two-sided learning, where both sides are uncertain. We propose a computationally efficient algorithm and prove that it asymptotically ($δ\to 0$) matches the lower bound to a constant for one-sided learning. Using the insights from the lower bound, we extend our algorithm to the two-sided learning setting and provide experimental results showing that it closely matches the lower bound on sample complexity. Finally, using a system of ODEs, we characterize the idealized fluid path that our algorithm chases.

   copy@inproceedings{pagare2025optimal,
     title     = {Optimal Algorithms for Bandit Learning in Matching Markets},
     author    = {Pagare, Tejas and Bandyopadhyay, Agniv and Juneja, Sandeep},
     booktitle = {NeurIPS Workshop on Dynamics at the Frontiers of Optimization, Sampling, and Games},
     year      = {2025}
   }

4. Explore-then-Commit Algorithms for Decentralized Two-Sided Matching MarketsTejas Pagare, Avishek Ghosh. IEEE International Symposium on Information Theory (ISIT 2024)
   [arXiv][pdf][slides][abstract][bibtex]

   Online learning in a decentralized two-sided matching markets, where the demand-side (players) compete to match with the supply-side (arms), has received substantial interest because it abstracts out the complex interactions in matching platforms (e.g. UpWork, TaskRabbit). However, past works assume that each arm knows their preference ranking over the players (one-sided learning), and each player aim to learn the preference over arms through successive interactions. Moreover, several (impractical) assumptions on the problem are usually made for theoretical tractability such as broadcast player-arm match Liu et al. (2020; 2021); Kong & Li (2023) or serial dictatorship Sankararaman et al. (2021); Basu et al. (2021); Ghosh et al. (2022). In this paper, we study a decentralized two-sided matching market, where we do not assume that the preference ranking over players are known to the arms apriori. Furthermore, we do not have any structural assumptions on the problem. We propose a multi-phase explore-then-commit type algorithm namely epoch-based CA-ETC (collision avoidance explore then commit) (\texttt{CA-ETC} in short) for this problem that does not require any communication across agents (players and arms) and hence decentralized. We show that for the initial epoch length of $T_{\circ}$ and subsequent epoch-lengths of $2^{l/γ} T_{\circ}$ (for the $l-$th epoch with $γ\in (0,1)$ as an input parameter to the algorithm), \texttt{CA-ETC} yields a player optimal expected regret of $\mathcal{O}\left(T_{\circ} (\frac{K \log T}{T_{\circ} Δ^2})^{1/γ} + T_{\circ} (\frac{T}{T_{\circ}})^γ\right)$ for the $i$-th player, where $T$ is the learning horizon, $K$ is the number of arms and $Δ$ is an appropriately defined problem gap. Furthermore, we propose a blackboard communication based baseline achieving logarithmic regret in $T$.

   copy@inproceedings{pagare2024explore,
     title     = {Explore-then-Commit Algorithms for Decentralized Two-Sided Matching Markets},
     author    = {Pagare, Tejas and Ghosh, Avishek},
     booktitle = {IEEE International Symposium on Information Theory (ISIT)},
     year      = {2024}
   }

5. Full Gradient Deep Reinforcement Learning for Average-Reward CriterionTejas Pagare, Vivek Borkar, Konstantin Avrachenkov. Learning for Dynamics and Control Conference (L4DC 2023)
   [arXiv][pdf][poster][abstract][bibtex]

   We extend the provably convergent Full Gradient DQN algorithm for discounted reward Markov decision processes from Avrachenkov et al. (2021) to average reward problems. We experimentally compare widely used RVI Q-Learning with recently proposed Differential Q-Learning in the neural function approximation setting with Full Gradient DQN and DQN. We also extend this to learn Whittle indices for Markovian restless multi-armed bandits. We observe a better convergence rate of the proposed Full Gradient variant across different tasks.

   copy@inproceedings{pagare2023full,
     title     = {Full Gradient Deep Reinforcement Learning for Average-Reward Criterion},
     author    = {Pagare, Tejas and Borkar, Vivek and Avrachenkov, Konstantin},
     booktitle = {Learning for Dynamics and Control Conference (L4DC)},
     year      = {2023}
   }

Master's Thesis

Talks & Posters

• slidesOptimal pure exploration algorithms for bandit learning in matching markets
• slidesTwo-sided bandit learning in matching marketsINRIA Sophia Antipolis
• slidesIntroduction to banditsAshoka University
• posterFull gradient reinforcement learning with average-reward criterionL4DC 2023

Gallery