On the Generation of Markov Decision Processes
By T. Archibald, K. McKinnon, Lyn C. Thomas. 1995
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Romain Laroche
- Jessica J. Chemali
- Layla El Asri
- Alessandro Lazaric
- Sergey Levine
- Bruno Scherrer
- Scott Fujimoto
- Aviral Kumar
Citations in the corpus
- Approximate Policy Iteration Schemes: A Comparison (2014, methodology): “More precisely, we consider Garnet problems first introduced by Archibald et al. (1995), which are a class of randomly constructed finite MDPs.”
- Off-policy learning with eligibility traces: a survey (2013, methodology): “More precisely, we consider Garnet problems (Archibald et al., 1995), which are a class of randomly constructed finite MDPs.”
- Score-based Inverse Reinforcement Learning (2016, background): “Here, we consider a special case of Garnets specified by three parameters: (NS , NA, NB).”
- Boosted and reward-regularized classification for apprenticeship learning (2014, methodology): “This experiment focuses on stationary Garnet problems, which are a class of randomly constructed finite MDPs representative of the kind of finite MDPs that might be encountered in practice [2].”
- Natural actor-critic algorithms (2009, unknown): “[6]).”
- Softened Approximate Policy Iteration for Markov Games (2016, unknown): “To empirically illustrate the application of quasi-Newton methods to both MDPs and MGs, we ran experiments on a class of randomly generated MDPs and MGs namely Garnets (see Archibald et al. (1995) for reference on Garnets on MDPs).”
- Learning from Demonstrations: Is It Worth Estimating a Reward Function? (2013, methodology): “The Garnet problems are a class of randomly constructed finite MDPs representative of the kind of finite MDPs that might be encountered in practice [2].”
- Direct Policy Iteration with Demonstrations (2015, background): “The Garnet framework [Archibald et al., 1995; Bhatnagar et al., 2009; Piot et al., 2014] allows to generate random finite MDPs characterized by 3 parameters: number of states N s , number of actions N a , and maximum branching factor N b .”
- Boosted Bellman Residual Minimization Handling Expert Demonstrations (2014, methodology): “A stationary Garnet problem is characterized by 3 parameters: Garnet ( N S , N A , N B ) .”
- Is the Bellman residual a bad proxy? (2016, unknown): “We consider Garnet problems [2, 4].”