Continuous probabilistic model building genetic network programming using reinforcement learning
Continuous probabilistic model building genetic network programming using reinforcement learning
复制标题
DOI：
10.1016/j.asoc.2014.10.023
复制
发表时间：
2015-02
期刊：
影响因子：
--
通讯作者：
Xianneng Li;K. Hirasawa
中科院分区：
文献类型：
--
作者：
Xianneng Li;K. Hirasawa
最近，一种新的概率建模进化算法(称为分布估计算法，简称EDA)被称为概率建模遗传网络规划(PMBGNP)。PMBGNP采用图结构进行个体表示，表现出比经典EDAS更强的表达能力。因此，它对EDAS进行了扩展，以解决数据挖掘和代理控制等一系列问题。针对智能体控制问题中的连续优化问题，提出了一种连续的PMBGNP算法。与其他连续进化算法不同，该算法通过强化学习(RL)来进化连续变量。在一个实际的移动机器人控制问题上，我们将其性能与几种最先进的算法进行了比较。结果表明，该算法的性能优于其他两种算法，具有统计学上的显著差异。
Recently, a novel probabilistic model-building evolutionary algorithm (so called estimation of distribution algorithm, or EDA), named probabilistic model building genetic network programming (PMBGNP), has been proposed. PMBGNP uses graph structures for its individual representation, which shows higher expression ability than the classical EDAs. Hence, it extends EDAs to solve a range of problems, such as data mining and agent control. This paper is dedicated to propose a continuous version of PMBGNP for continuous optimization in agent control problems. Different from the other continuous EDAs, the proposed algorithm evolves the continuous variables by reinforcement learning (RL). We compare the performance with several state-of-the-art algorithms on a real mobile robot control problem. The results show that the proposed algorithm outperforms the others with statistically significant differences.