Parametric Bandits: The Generalized Linear Case

By Sarah Filippi, Olivier Capp'e, Aurélien Garivier, Csaba Szepesvari. 2010

529 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.

Citations in the corpus

Weight Uncertainty in Neural Networks (2015, background): “Contextual bandits are simple reinforcement learning problems without persistent state (Li et al., 2010; Filippi et al., 2010).”
Off-policy evaluation for slate recommendation (2016, background): “, linear) models relating the metrics to the features describing a slate [2, 31, 15, 10, 29]; this can lead to bias if the model is inaccurate (e.”
Unbiased offline evaluation of contextual-bandit-based news article recommendation algorithms (2010, background): “Extensions to generalized linear models [12] are also possible and can still enjoy the same ˜ O ( √ T ) regret guarantee.”
Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems (2012, positive): “[78] extends the analysis of fixed design by assuming a generalized linear model to capture the dependence of rewards on covariates.”
Learning to Optimize via Posterior Sampling (2013, background): “[18] can be applied to establish a Bayesian regret bound for posterior sampling, but with one caveat.”
Thompson Sampling for Contextual Bandits with Linear Payoffs (2012, background): “At time t = 1 , 2 , . . . , a context vector b i ( t ) ∈ R d , is revealed for every arm i .”
On the Theory of Reinforcement Learning with Once-per-Episode Feedback (2021, methodology): “We note that such boundedness assumptions are standard in the logistics bandits literature [13, 31, 14].”
Optimal Algorithms for Stochastic Contextual Preference Bandits (2021, methodology): “Sequential decision-making problems with side information (in the form of features or attributes), have been popular in machine learning as contextual bandits [16, 13, 26].”
Returning is Believing: Optimizing Long-term User Engagement in Recommender Systems (2017, methodology): “e explicit modeling of user return dierentiates r2Bandit from existing bandit algorithms for online recommendation [12, 20].”
Interactive collaborative filtering (2013, methodology): “Using the conclusion from [14], GLM-UCB has a regret bound”
Neural Contextual Bandits with UCB-based Exploration (2019, background): “…propose a neural contextual bandit algorithm that can be regarded as an extension of existing (generalized) linear bandit algorithms (Abbasi-Yadkori et al., 2011; Filippi et al., 2010; Li et al., 2010; 2017) to the case of arbitrary bounded reward functions. i and its j -th coordinate by [ θ ] j .”
Learning Hidden Features for Contextual Bandits (2016, methodology): “, LinUCB [15] and GLM-UCB [10], we can find that our algorithm considers not only the prediction confidence of user preference parameters (the second term), but also the confidence of learnt hidden features (the third term).”
Bandit Theory meets Compressed Sensing for high dimensional Stochastic Linear Bandit (2012, methodology): “On the contrary, Linear Bandit Theory (see e.g. Rusmevichientong and Tsitsiklis (2008); Dani et al. (2008); Filippi et al. (2010) and the recent work by Abbasi-yadkori et al. (2011)) addresses this issue of maximizing the sum of rewards by eﬃciently balancing between exploration and exploitation.”
Optimistic Bayesian Sampling in Contextual-Bandit Problems (2012, methodology): “Our definition differs slightly from that of Filippi et al. (2010) since we do not restrict attention to generalised linear bandits.”
Reinforcement Learning with Trajectory Feedback (2020, background): “…we believe that the results can be adapted to cases where the feedback is a more complex mapping from state-actions into trajectory-reward, and, specifically, a noisy generalized linear model (GLM) of the trajectory (Filippi et al. 2010; Abeille, Lazaric et al. 2017; Kveton et al. 2020).”
Provable Optimal Algorithms for Generalized Linear Contextual Bandits (2017, methodology): “Recently, Filippi et al. (2010) study the same stochastic GLM bandit problem considered here.”
Contextual Bandits with Latent Confounders: An NMF Approach (2016, background): “In [15] the authors have expanded this model for the generalized linear model regime.”