Battle of Bandits

By Aadirupa Saha, Aditya Gopalan. 2018

48 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Reinforcement Learning

Citations in the corpus

Dueling RL: Reinforcement Learning with Trajectory Preferences (2021, methodology): “Yue and Joachims (2009, 2011); Saha and Krishnamurthy (2022); Ghoshal and Saha (2022); Saha and Gopalan (2018a) The setting is relevant in various realworld systems which aim to collect information from user preferences, including recommender systems, retail management, search engine optimization,…”
Exploiting Correlation to Achieve Faster Learning Rates in Low-Rank Preference Bandits (2022, methodology): “The above algorithm exhibits improved sample complexity O(rε−2 log(n/δ)) for all ε ∈ (0, (r/n3)1/2]; this improves on the O(nε−2 log(n/δ)) of Saha and Gopalan (2020) for the I-RUM(n, k) model. In particular, we don’t need n to be constant for the overall sample complexity to be O(rε−2 log(rδ)) i.e.,”
Optimal Algorithms for Stochastic Contextual Preference Bandits (2021, background): “Above notion of learner’s regret is motivated from the definition of classical K-armed dueling bandit regret introduced by [44] which is later adopted by the dueling bandit literature [46, 23, 3, 43, 45, 38, 31].”
Principled Reinforcement Learning with Human Feedback from Pairwise or K-wise Comparisons (2023, background): “…on pairwise comparisons (Yue et al., 2012; Zoghi et al., 2014b; Yue and Joachims, 2009, 2011; Saha and Krishnamurthy, 2022; Ghoshal and Saha, 2022; Saha and Gopalan, 2018a; Ailon et al., 2014; Zoghi et al., 2014a; Komiyama et al., 2015; Gajane et al., 2015; Saha and Gopalan, 2018b, 2019; Faury et…”
PAC Battling Bandits in the Plackett-Luce Model (2018, background): “The battling bandit decision framework (Saha and Gopalan, 2018; Chen et al., 2018) models several application domains where it is possible to elicit feedback about preferred options from among a general set of oﬀered options, instead of being able to compare only two options at a time as in the…”
MergeDTS (2018, background): “sely, the cumulative regret of MergeRUCB is almost twice as large as that of MergeDTS in the large-scale setup shown in Section6. A recent extension of dueling bandits is called multi-dueling bandits [3, 33, 35], where more than two rankers can be compared at each step. Multi-Dueling Bandit (MDB) is”