Non-parametric policy gradients: a unified treatment of propositional and relational domains

By Kristian Kersting, Kurt Driessens. 2008

76 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 Artificial Intelligence

Citations in the corpus

Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, background): “Typically, in relational reinforcement learning, policies are represented as relational regression trees [90, 91, 143, 96, 76] that are more interpretable than neural networks.”
Fitted Q-Learning for Relational Domains (2020, methodology): “However, in relational domains this information is almost never available: without dif-ﬁcult feature engineering, the shape of the value function is almost always unknown – a reason why some work directly operated at the policy level (Fern, Yoon, and Givan 2006; Kersting and Driessens 2008).”
Ieee Transactions on Systems, Man and Cybernetics—part C: Applications and Reviews 1 a Survey of Actor-critic Reinforcement Learning: Standard and Natural Policy Gradients (unknown): “Although not explicitly considered here, work on nonparametric policy gradients does exist, see e.g. [34], [35].”
Learning Markov Logic Networks via Functional Gradient Boosting (2011, background): “The functional gradient approach has produced state of the art results in building relational dependency networks over many domains [14] and has been employed in other relational learning problems such as relational CRFs [15], relational policies [16], relational sequences [17], etc.”
Human-Guided Learning for Probabilistic Logic Models (2018, methodology): “This method has been extended to various relational models for learning the structure (Kersting and Driessens, 2008; Karwath et al., 2008; Natarajan et al., 2011; Natarajan et al., 2012).”
Solving Relational and First-Order Logical Markov Decision Processes: A Survey (2012, background): “Recently, in the non-parametric policy gradient ( NPPG) approach, Kersting and Driessens (2008) applied the idea of gradient boosting.”
Gradient-based boosting for statistical relational learning: The relational dependency network case (2011, methodology): “2000) and their relational extensions (Kersting and Driessens 2008).”
Structure learning for relational logistic regression: an ensemble approach (2018, methodology): “Inspired by the recent successes of non-parametric learning methods for SRL models (Gutmann and Kersting 2006; Kersting and Driessens 2008; Natarajan et al. 2017, 2011, 2012), we develop a non-parametric learning method for full model learning (structure and parameters) of RLR models.”
Statistical Relational Artificial Intelligence: Logic, Probability, and Computation (2016, methodology): “is corresponds to a functional gradient approach and was originally proposed for training conditional random fields for labeling relational sequences [Gutmann and Kersting, 2006] and for learning relational policies [Kersting and Driessens, 2008].”
Policy Tree: Adaptive Representation for Policy Gradient (2015, background): “One notable work on adaptive representation for policy gradient includes the NPPG algorithm (Kersting and Driessens 2008).”