Learning Policy Improvements with Path Integrals

69 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.

Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Review (2018, methodology): “The path integral framework has been used to derive algorithms for both optimal control and planning (Kappen, 2011) and policy search via reinforcement learning (Theodorou et al., 2010).”
Learning dexterous manipulation for a soft robotic hand from human demonstrations (2016, methodology): “A widely used approach is to learn the parameters of a dynamic motor primitive [10] (DMP) with relative entropy policy search [11] or PI2 [12].”
Black-Box Policy Search with Probabilistic Programs (2015, background): “ally weighted cost-to-go is known as risk-sensitive optimal control [van den Broek et al., 2010]. Path integral approaches have been applied to reinforcement learning problems in a number of studies [Theodorou et al., 2007, Theodorou and Todorov, 2012, Azar and Kappen, 2012]. Related approaches [Raw”
Combining the benefits of function approximation and trajectory optimization (2014, unknown): “Such optimization is rarely used on real robots, even though model-free methods exist for adapting a given trajectory to unknown dynamics [34].”
Perception, Action, and Utility: The Tangled Skein (2012, background): “Hence, this pattern of preferences is sometimes known as ambiguity aversion, and has been repeatedly demonstrated in a variety of paradigms (see Loewenstein et al., 2008, for a recent review).”