Program evolution of genetic programming based on probabilistic grammar

在本研究中，我们提出了“基于概率语法的程序进化”，并通过扩展遗传规划（GP）构造了自动规划方法。在这种基于种群的方法中，我们不使用遗传操作符，如交叉或突变。相反，我们估计概率语法的概率分布，根据该概率语法生成程序。我们已经通过应用于实际问题验证了所提出的方法，例如，设计自动化、机器人、金融工程和Web智能等。

In this research, we have proposed the “program evolution based on the probabilistic grammar” and constructed the automatic programming method by extending Genetic Programming (GP). In this population-based method, we do not employ genetic operators, such as crossovers or mutations. Instead, we estimate the probabilistic distribution of a probabilistic grammar, according to which a program is generated. We have validated the proposed method by applying to real-world problems, e.g., design automation, robotics, financial engineering, and Web intelligence etc.