THE ROLE OF EXPLORATION IN LEARNING CONTROL

By D. Sofge. 1992

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

Self-Imitation Learning (2018, background): “Though the role of exploitation for exploration has been discussed (Thrun, 1992), prior work has mostly considered exploiting what the agent has learned, whereas we consider exploiting what the agent has experienced, but has not yet learned.”
An empirical evaluation of interval estimation for Markov decision processes (2004, background): “Many approaches to exploration in MDPs have been proposed (Thrun, 1992; Wiering & Schmidhuber, 1998; Dearden et al., 1999; Meuleau & Bourgine, 1999; Wyatt, 2001).”
Bayesian Q-Learning (1998, background): “For a survey of directed and undirected exploration techniques, see [13].”
Exploration Bonuses and Dual Control (1996, methodology): “The agent could use these uncertainties in an analogue of the systems of Schmidhuber (1991) or Thrun & M ̈ oller (1992). The latter make the agent learn the areas of the state space in which it makes errors in its predictions (of the value or transition functions of the environment); however, they a”
Reinforcement Learning and Dynamic Programming Using Function Approximators (2010, background): “Hence, thi s second requirement is partly in conflict with the first, and the combination of the tw o is traditionally called the exploration-exploitation trade-off in the RL literatu re (Thrun, 1992; Kaelbling, 1993; Sutton and Barto, 1998).”
Point-based value iteration: an anytime algorithm for POMDPs (2003, methodology): “Finally, exploiting an insight from policy search methods and MDP exploration [Ng and Jordan, 2000; Thrun, 1992], PBVI uses explorative stochastic trajectories to select belief points, thus reducing the number of belief points necessary to ﬁnd a good solution.”
Modeling functions of striatal dopamine modulation in learning and planning. (2001, methodology): “In reinforcement learning studies, one popular technique is to be optimistic for places that have never been explored (Thrun 1992; Dayan and Sejnowski 1996; Sutton and Barto, 1998).”
Sample Complexity Bounds of Exploration (2012, background): “A number of alternatives such as soft-max, counter-based, recency-based, optimistic initialization, and exploration bonus are sometimes found more effective empirically (Sutton and Barto, 1998; Thrun, 1992).”
Near-Optimal Reinforcement Learning in Polynomial Time (2002, background): “Thus, the reinforcement learning agent faces a fundamental trade-o between exploitation and exploration (Thrun, 1992; Sutton & Barto, 1998): should the agent exploit its cumulative experience so far, by executing the action that currently seems best, or should it execute a di erent action, with the ”
Optimal control theory and the linear bellman equation (2011, methodology): “Well-known examples of this appraoch are Bayesian reinforcement learning Dearden et al. (1999) and some older methods that are reviewd in Thrun (1992).”
Reinforcement Learning: A Tutorial. (1997, unknown): “For a good introduction to the issues of efficient exploration see Thrun (1992).”