Analysis and Optimization of Deep Counterfactual Value Networks
Artificial Intelligence
2018-10-15 v2
Abstract
Recently a strong poker-playing algorithm called DeepStack was published, which is able to find an approximate Nash equilibrium during gameplay by using heuristic values of future states predicted by deep neural networks. This paper analyzes new ways of encoding the inputs and outputs of DeepStack's deep counterfactual value networks based on traditional abstraction techniques, as well as an unabstracted encoding, which was able to increase the network's accuracy.
Keywords
Cite
@article{arxiv.1807.00900,
title = {Analysis and Optimization of Deep Counterfactual Value Networks},
author = {Patryk Hopner and Eneldo Loza Mencía},
journal= {arXiv preprint arXiv:1807.00900},
year = {2018}
}
Comments
Long version of publication appearing at KI 2018: The 41st German Conference on Artificial Intelligence (http://dx.doi.org/10.1007/978-3-030-00111-7_26). Corrected typo in title