Approximate Dynamic Programming Finally Performs Well in the Game of Tetris
By Victor Gabillon, Mohammad Ghavamzadeh, Bruno Scherrer. 2013
60 citations recorded by the source corpus.
Subfields: Reinforcement Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Reinforcement Learning
- Sergey Levine
- Pieter Abbeel
- Shimon Whiteson
- Ofir Nachum
- Chelsea Finn
- Shie Mannor
- Aviral Kumar
- Matthieu Geist
Citations in the corpus
- Trust Region Policy Optimization (2015, methodology): “(See [5] for a review of this domain and the strongest counterattack so far from the ADP side.”
- Generalization and Exploration via Randomized Value Functions (2014, background): “Tetris has been something of a benchmark problem for RL and approximate dynamic programming, with several papers on this topic (Gabillon et al., 2013).”
- Analysis of Classification-based Policy Iteration Algorithms (2016, background): “2013) and it has been shown to be empirically competitive in a series of testbeds and challenging applications (see e.g., Farahmand et al. 2013; Gabillon et al. 2013).”
- Control of a Quadrotor With Reinforcement Learning (2017, methodology): “We consider a simple exploration strategy that is described in [9], [10] as shown in Fig.”
- Policy Gradients Beyond Expectations: Conditional Value-at-Risk (2014, methodology): “Many approaches to learning controllers for Tetris are described in the literature, among them are approximate value iteration (Tsitsiklis & Van Roy, 1996), policy gradients (Kakade, 2001), and modified policy iteration (Gabillon et al., 2013).”
- Optimizing the CVaR via Sampling (2014, methodology): “We do point out, however, that whether the approach of Gabillon, Ghavamzadeh, and Scherrer (2013) could be extended to handle a CVaR objec-tive is currently not known.”
- Approximate modified policy iteration and its application to the game of Tetris (2015, background): “This paper is a significant extension of two conference papers by the authors (Scherrer et al., 2012; Gabillon et al., 2013).”
- Online meta-learning by parallel algorithm competition (2017, methodology): “The current state-of-the-art result for a single run of an algorithm, achieved by the CBMPI algorithm [12], is a mean score of 51 million cleared lines.”