Concurrent Reinforcement Learning from Customer Interactions

By David Silver, Leonard Newnham, David Barker, Suzanne Weller, Jason McFall. 2013

72 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

Distributed Prioritized Experience Replay (2018, background): “This includes simulated environments but also a variety of real-world applications, such as robotic arm farms, self-driving cars, online recommender systems, or other multi-user systems in which data is generated by many instances of the same environment (c.f. Silver et al., 2013).”
Massively Parallel Methods for Deep Reinforcement Learning (2015, background): “Another paradigm which has been explored is concurrent reinforcement learning (Silver et al., 2013), in which an agent can interact in parallel with an inherently distributed environment, e.g. to optimize interactions with multiple users on the internet.”
Recurrent Reinforcement Learning: A Hybrid Approach (2015, methodology): “Later, Silver et al. (2013) apply variants of Q-learning to learn a linear Q-function in the CRM task of email campaigns.”
Safe Reinforcement Learning (2015, background): “For simplicity, we do not consider issues that arise due to the parallel and continuing nature of the problem—we ignore that trajectories occur in parallel (Silver et al., 2013) and that there is no way to determine whether a trajectory has ended (users may return after an arbitrarily long hiatus).”
Concurrent Meta Reinforcement Learning (2019, unknown): “The concurrent setting is deﬁned as the case where episodic interactions occur in parallel, and Silver et al. (2013) 2016) had a dedicated communication channel between two agents that was either discrete or continuous, with potential channel noise added.”
Exploring or Exploiting? Social and Ethical Implications of Autonomous Experimentation in AI (2016, unknown): “Here, for example, the environment might be user attributes, the actions might be be various ad placements, and the reward might be increasing revenue [5, 17].”
Mastering the game of Go without human knowledge (2017, background): “However, very similar algorithms have subsequently proven highly effective in video games 6–8,10 , robotics 60 , industrial control 61–63 and online recommendation systems 64,65 .”
A Batch, Off-Policy, Actor-Critic Algorithm for Optimizing the Average Reward (2016, background): “In contrast, Silver et al. (2013) developed a on-policy, online temporal-difference algorithm for learning a policy based on sequences of interactions with multiple individuals.”
Concurrent PAC RL (2015, background): “One important exception is recent work by Silver et al. (2013) on concurrent reinforcement learning when interacting with a set of customers in parallel.”
Deep Reinforcement Learning: An Overview (2017, background): “Silver et al. (2013) proposed concurrent reinforcement learning for the customer interaction problem.”
Coordinated Exploration in Concurrent Reinforcement Learning (2018, background): “In concurrent reinforcement learning (Silver et al., 2013; Pazis & Parr, 2013; Guo & Brunskill, 2015; Pazis & Parr, 2016), multiple agents interact with different instances of the same unknown environment, have perfect information sharing with one another, and learn in parallel how to achieve a comm”
Contextual Markov Decision Processes (2015, background): “The concurrent RL setup (Silver et al., 2013), as well as the case of many or even inﬁnitely many contexts are of practical importance.”