arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1807.00900cs.AI
收藏

深度虚拟价值网络的分析与优化

Analysis and Optimization of Deep Counterfactual Value Networks

Technische Universität Darmstadt(达姆施塔特工业大学)

Patryk Hopner, Eneldo Loza Mencía

摘要:

最近发布了一种名为DeepStack的强大扑克算法,该算法能够通过使用深度神经网络预测的未来状态的启发式值来在游戏过程中找到近似的纳什均衡。本文分析了基于传统抽象技术对DeepStack深度反事实价值网络的输入和输出进行编码的新方法,以及能够提高网络准确性的非抽象编码。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑