DEVELOPING A PREDICTIVE APPROACH TO KNOWLEDGE

By Adam White. 2015

70 citations recorded by the source corpus.

Subfields: Computer Science Applications

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Science Applications

Citations in the corpus

An Off-policy Policy Gradient Theorem Using Emphatic Weightings (2018, background): “Most importantly, off-policy methods allow an agent to learn about many different policies at once, forming the basis for a predictive understanding of an agent’s environment [Sutton et al., 2011, White, 2015] and enabling the learning of options [Sutton et al., 1999, Precup, 2000].”
Unifying Task Specification in Reinforcement Learning (2016, background): “The separation into underlying MDP dynamics and task speciﬁcation is particularly useful in off-policy learning, with the Horde formalism, where many demons (value functions) are learned off-policy.”
True Online Emphatic TD(λ): Quick Reference and Implementation Guide (2015, background): “We consider the setting of general value functions, or GVFs (Maei & Sutton 2010, Sutton et al. 2011, White 2015, Sutton, Mahmood & White 2015).”
Online Off-policy Prediction (2018, background): “In practice, the best values of α and αh are problem dependent, and the practitioner must tune them independently to achieve good performance (White, 2015; White & White, 2016).”
Multi-step Off-policy Learning Without Importance Sampling Ratios (2017, methodology): “Learning a large number of oﬀ-policy predictions is also considered important for model learning, options learning (Sutton et al. 1999), scalable life-long learning (White, Modayil & Sutton 2012), and knowledge representation (White 2015).”
A Greedy Approach to Adapting the Trace Parameter for Temporal Difference Learning (2016, background): “, learning from demonstrations), large importance sampling ratios can destabilize provably convergent gradient temporal difference learning methods, when the parameters are not set in a very particular way (λ = 0) [33].”
Planning with Expectation Models (2019, background): “Policy evaluation is also key for building predictive world knowledge where the questions about the world are formulated using value functions [Sutton et al. , 2011; Modayil et al. , 2014; White and others, 2015].”
Discovery of Useful Questions as Auxiliary Tasks (2019, methodology): “We chose the GVF formulation for its flexibility: according to the reward hypothesis (Sutton & Barto, 2018), any goal might be formulated in terms of a scalar signal, or cumulant (White, 2015), whose discounted sum must be maximized.”
Investigating Practical Linear Temporal Difference Learning (2016, background): “We intentionally precluded smaller values of αh from the parameter sweep because many of the gradient TD methods simply become their on-policy variants as αh approaches zero, whereas in some off-policy domains values of αh > α are required to avoid divergence [26].”
Incremental Off-policy Reinforcement Learning Algorithms (2017, unknown): “White (2015) comprehensively explored this framework.”