Author name cluster

Krishna Acharya

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

3 papers

2 author rows

TMLR Journal 2025 Journal Article

Producers Equilibria and Dynamics in Engagement-Driven Recommender Systems

Krishna Acharya
Varun Vangala
Jingyan Wang
Juba Ziani

Online platforms such as YouTube, Instagram heavily rely on recommender systems to decide what content to present to users. Producers, in turn, often create content that is likely to be recommended to users and have users engage with it. To do so, producers try to align their content with the preferences of their targeted user base. In this work, we explore the equilibrium behavior of producers who are interested in maximizing user engagement. We study two variants of the content-serving rule for the platform's recommender system, and provide a structural characterization of producer behavior at equilibrium: namely, each producer chooses to focus on a single embedded feature. We further show that specialization, defined as different producers optimizing for distinct types of content, naturally emerges from the competition among producers trying to maximize user engagement. We provide a heuristic for computing equilibria of our engagement game, and evaluate it experimentally. We highlight i) the performance and convergence of our heuristic, ii) the degree of producer specialization, and iii) the impact of the content-serving rule on producer and user utilities at equilibrium and provide guidance on how to set the content-serving rule.

PDF Details

UAI Conference 2024 Conference Paper

One Shot Inverse Reinforcement Learning for Stochastic Linear Bandits

Etash Guha
Jim James
Krishna Acharya
Vidya Muthukumar
Ashwin Pananjady

The paradigm of inverse reinforcement learning (IRL) is used to specify the reward function of an agent purely from its actions and is critical for value alignment and AI safety. While IRL is successful in practice, theoretical guarantees remain nascent. Motivated by the need for IRL in large action spaces with limited data, we consider as a first step the problem of learning from a single sequence of actions (i. e. , a demonstration) of a stochastic linear bandit algorithm. When the demonstrator employs the Phased Elimination algorithm, we develop a simple inverse learning procedure that estimates the linear reward function consistently in the time horizon with just a single demonstration. In particular, we show that our inverse learner approximates the true reward parameter within a error of $\mathcal{O}(T^{-\frac{\omega - 1}{2\omega }})$ (where $T$ is the length of the demonstrator’s trajectory and $\omega$ is a constant that depends on the geometry of the action set). We complement this result with an information-theoretic lower bound for any inverse learning procedure. We corroborate our theoretical results with simulations on synthetic data and a demonstration constructed from the MovieLens dataset.

Details

ICLR Conference 2024 Conference Paper

Oracle Efficient Algorithms for Groupwise Regret

Krishna Acharya
Eshwar Ram Arunachaleswaran
Sampath Kannan
Aaron Roth 0001
Juba Ziani

We study the problem of online prediction, in which at each time step $t \in \{1,2, \cdots T\}$, an individual $x_t$ arrives, whose label we must predict. Each individual is associated with various groups, defined based on their features such as age, sex, race etc., which may intersect. Our goal is to make predictions that have regret guarantees not just overall but also simultaneously on each sub-sequence comprised of the members of any single group. Previous work such as [Blum & Lykouris][1] and [Lee et al][2] provide attractive regret guarantees for these problems; however, these are computationally intractable on large model classes (e.g., the set of all linear models, as used in linear regression). We show that a simple modification of the sleeping experts technique of [Blum & Lykouris][1] yields an efficient *reduction* to the well-understood problem of obtaining diminishing external regret *absent group considerations*. Our approach gives similar regret guarantees compared to [Blum & Lykouris][1]; however, we run in time linear in the number of groups, and are oracle-efficient in the hypothesis class. This in particular implies that our algorithm is efficient whenever the number of groups is polynomially bounded and the external-regret problem can be solved efficiently, an improvement on [Blum & Lykouris][1]'s stronger condition that the model class must be small. Our approach can handle online linear regression and online combinatorial optimization problems like online shortest paths. Beyond providing theoretical regret bounds, we evaluate this algorithm with an extensive set of experiments on synthetic data and on two real data sets --- Medical costs and the Adult income dataset, both instantiated with intersecting groups defined in terms of race, sex, and other demographic characteristics. We find that uniformly across groups, our algorithm gives substantial error improvements compared to running a standard online linear regression algorithm with no groupwise regret guarantees.

Details