Multi-Bandit Best Arm Identification

By Victor Gabillon, Mohammad Ghavamzadeh, Alessandro Lazaric, Sébastien Bubeck. 2011

123 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

Taking the Human Out of the Loop: A Review of Bayesian Optimization (2016, background): “There are many results on exploration [30], [51], [50] and contextual bandits [97], [112], [2].”
Learning to Learn for Global Optimization of Black Box Functions (2016, methodology): “We are therefore considering the ﬁnite horizon setting that is popular in AB tests Kohavi et al. [2009], Scott [2010] and is often studied under the umbrella of best arm identiﬁcation in the bandits literature [Bubeck et al., 2009, Gabillon et al., 2011, 2012, Hoffman et al., 2014].”
Almost Optimal Exploration in Multi-Armed Bandits (2013, background): “Another related setting was introduced by Gabillon et al. (2011), who consider a multi-bandit MAB problem in which each player has to identify the best arm in his sub-problem and the challenge is to distribute the limited resources at hand (namely T arm pulls) so as to maximize the confidence of…”
Best Arm Identification: A Unified Approach to Fixed Budget and Fixed Confidence (2012, background): “, Bernstein-based bounds) and to more complex settings, such as the multi-bandit best arm identification problem introduced in [8].”
Multiple Identifications in Multi-Armed Bandits (2012, methodology): “Finally we consider the extension of UCB-E to the m-best arms identification problem, which is based on a similar idea than the extension Gap-E of (Gabillon et al., 2011) for the multi-bandit best arm identification, see Figure 3 for the details.”
On correlation and budget constraints in model-based bandit optimization with application to automatic machine learning (2014, background): “In this same vein, it is important to take into account that a recommendation must be made at time T in order to properly allocate actions and explore the space of possible optima.”
Budgeted Learning for Developing Personalized Treatment (2014, background): “Several frequentist’s approaches with a similar objective have been studied in [13]–[15].”
The strategic student approach for life-long exploration and learning (2012, background): “In this setting the learner is tested after a learning period and it has either to declare what is the best arm [34]”
Learning to Hire Teams (2015, methodology): “Next, we extend the algorithms to multiple types of tasks (M > 1) by jointly learning over all of these task types, and adaptively allocating the budget across them capturing the inherent hardness of best action selection problem per task type (Gabillon et al. 2011).”