Corpus ID: 15984219

Applying Q-Learning to Flappy Bird

@inproceedings{EbelingRump2016ApplyingQT,
  title={Applying Q-Learning to Flappy Bird},
  author={Moritz Ebeling-Rump and Zachary Hervieux-Moore},
  year={2016}
}
The field of machine learning is an interesting and relatively new area of research in artificial intelligence. In this paper, a special type of reinforcement learning, Q-Learning, was applied to the popular mobile game Flappy Bird. The QLearning algorithm was tested on two different environments. The original version and a simplified version. The maximum score achieved on the original version and simplified version were 169 and 28,851, respectively. The trade-off between runtime and accuracy… Expand

Figures from this paper

REINFORCEMENT LEARNING ALGORITHMS IN THE COMPUTER GAME FLAPPY BIRD
Nowadays, machine learning and artificial intelligence have become very attractive because of its ability, sometimes more and sometimes less, to provide an almost optimal solution to problems whereExpand
Playing Flappy Bird via Asynchronous Advantage Actor Critic Algorithm
TLDR
In this study, the model was trained with raw game images, which had not been seen before, and the trained model has learned as reinforcement when to make which decision. Expand
A Minimal Training Strategy to Play Flappy Bird Indefinitely with NEAT
TLDR
A minimal training strategy to develop autonomous virtual players using the NEAT neuroevolutionary algorithm to evolve an agent capable of playing the Flappy Bird game is proposed. Expand

References

SHOWING 1-6 OF 6 REFERENCES
Reinforcement Learning: An Introduction
TLDR
This book provides a clear and simple account of the key ideas and algorithms of reinforcement learning, which ranges from the history of the field's intellectual foundations to the most recent developments and applications. Expand
Q-learning
TLDR
This paper presents and proves in detail a convergence theorem forQ-learning based on that outlined in Watkins (1989), showing that Q-learning converges to the optimum action-values with probability 1 so long as all actions are repeatedly sampled in all states and the action- values are represented discretely. Expand
Asynchronous Stochastic Approximation and Q-Learning
TLDR
The Q-learning algorithm, a reinforcement learning method for solving Markov decision problems, is studied to establish its convergence under conditions more general than previously available. Expand
Learning from delayed rewards
  • B. Kröse
  • Computer Science
  • Robotics Auton. Syst.
  • 1995
The invention relates to a circuit for use in a receiver which can receive two-tone/stereo signals. This circuit is intended to make a choice between mono or stereo reproduction of signal A or ofExpand
Everything you need to know about your new favorite cell phone game, 'flappy bird
  • Everything you need to know about your new favorite cell phone game, 'flappy bird
  • 2014
What is flappy bird? the game taking the app store by storm
  • What is flappy bird? the game taking the app store by storm
  • 2014