Learning predictive state representations using non-blind policies
By Michael H. Bowling, Peter Mccracken, Michael R. James, James Neufeld, Dana F. Wilkinson. 2006
52 citations recorded by the source corpus.
Subfields: Artificial Intelligence, Control and Systems Engineering
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Sergey Levine
- Byron Boots
- Dieter Fox
- Chelsea Finn
- Geoffrey J. Gordon
- Arthur Gretton
- Arunkumar Byravan
- Pieter Abbeel
In Control and Systems Engineering
- Dieter Fox
- Arunkumar Byravan
- Byron Boots
- Sergey Levine
- Geoffrey J. Gordon
- Arthur Gretton
- Chelsea Finn
- Marc Peter Deisenroth
Citations in the corpus
- Hilbert Space Embeddings of Predictive State Representations (2013, background): “In this paper we restrict ourselves to linear PSRs, in which all prediction functions are linear: fτl(s(ht)) = (1)For simplicity, we assume that all probabilities involving actions refer to our PSR as controlled by an arbitrary blind or open-loop policy [15] (also called exogenous inputs).”
- An Online Spectral Learning Algorithm for Partially Observable Nonlinear Dynamical Systems (2011, methodology): “We can compute all elements of Σ̂T ,H from a single sample of trajectories if we sample histories from ω, follow an appropriate exploratory policy, and then importanceweight each sample (Bowling et al. 2006): [Σ̂T ,H]ij is∑w t=1 λtφ T itφ H jt, where λt is an importance weight.”
- Learning predictive models of a depth camera & manipulator from raw execution traces (2014, background): “2For simplicity, we assume that all probabilities involving actions refer to our PSR as controlled by an arbitrary blind or open-loop policy [31].”
- Self-Correcting Models for Model-Based Reinforcement Learning (2016, unknown): “We now focus on blind policies (Bowling et al. 2006).”