Bandit Problems: Sequential Allocation of Experiments.

By J. Bather, D. Berry, B. Fristedt. 1986

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

A contextual-bandit approach to personalized news article recommendation (2010, background): “The context-free K-armed bandit problem has been studied by statisticians for a long time [9, 24, 26].”
Unbiased offline evaluation of contextual-bandit-based news article recommendation algorithms (2010, background): “The multi-armed bandit problem [7] is a classic and popular model for studying the exploration-exploitation tradeoﬀ.”
Reinforcement Learning with a Corrupted Reward Channel (2017, background): “Previous related works have mainly focused on the wireheading case of Example 3 [Bostrom, 2014; Yampolskiy, 2014], also known as self-delusion [Ring and Orseau, 2011], and reward hacking [Hutter, 2005, p. 239]. A notable exception is Amodei et al. [2016], who argue that corrupt reward is not limited”
Exploration of Multi-State Environments: Local Measures and Back-Propagation of Uncertainty (1999, background): “The Berry and Fristedt book (1985) provides an extensive bibliography on the subject.”
Cognitive Medium Access: Exploration, Exploitation, and Competition (2007, background): “The key observation motivat ing our approach is the equivalence between our problem and the classical multi-armed bandit problem (see [5] and references therein).”
Personalizing persuasive technologies: Explicit and implicit personalization using persuasion profiles (2015, background): “The exploration–exploitation problem is broadly studied under the heading of the multi-armed bandit problem (see, e.g., Whittle, 1980; Berry and Fristedt, 1985; Sutton and Barto, 1998).”
An unbiased offline evaluation of contextual bandit algorithms with generalized linear models (2011, background): “The multi-armed bandit problem is a classic and popular model for studying the explorationexploitation tradeoff (Berry and Fristedt, 1985).”
Rationality and Intelligence: A Brief Update (2013, background): “It views the metalevel decision problem as a multi-armed bandit problem (Berry and Fristedt, 1985) and applies an asymptotically near-optimal bandit decision rule recursively to make a choice of which computation to do next.”
Continuously adaptive continuous queries over streams (2002, background): “The adaptivity of Eddies relates them closely to work in control theory [7, 10, 5, 23, 30] and reinforcement learning [36], and we are interested in extending eddy routing policies to adaptively learn data dependencies and query dependencies in a more sophisticated manner than typically done in data”
Decision theory, reinforcement learning, and the brain (2008, methodology): “Problems of this sort are surprisingly computationally intractable (Gittins, 1989; Berry and Fristedt, 1985); section 3.4 discusses the issues and approximate solutions, including one that, evidence suggests, implicates fronto-polar cortex.”
Multi-armed Bandit Models for the Optimal Design of Clinical Trials: Benefits and Challenges (2015, methodology): “The calibration method uses DP to approximate the Gittins index values based on this idea, as explained in Gittins and Jones (1979), and it can be adapted to compute the ﬁnite-horizon counterpart, as explained in Berry and Fristedt (1985), Chapter 5.”
Design Issues for Generalized Linear Models: A Review (2006, background): “Freeman (1970), Owen (1975), Kuo (1983) and Berry and Fristedt (1985) have also done work in the sequential domain.”
RANDOMIZED ALLOCATION WITH NONPARAMETRIC ESTIMATION FOR A MULTI-ARMED BANDIT PROBLEM WITH COVARIATES (2002, background): “Readers are referred to Berry and Fristedt (1985) and Gittins (1989) and the references cited therein for the history, many elegant results and applications in clinical trials, scheduling and other industrial problems.”