基于分布式强化学习算法的精准助学数据分析方法研究

Design of the data analysis method of accurate financial aid data recognition based on distributed reinforcement learning algorithm

针对传统助学金评选方法中存在虚假申请材料难以鉴别、无法准确了解学生真实经济水平的问题。文中提出了基于分布式强化学习算法的精准助学数据分析方法,该方法通过收集数字化校园中学生的各项消费数据,经过数据预处理后找出具有分类能力的变量。然后,将RBF神经网络通过归一化与选取合适的隐藏层层数、神经元个数来提高聚类速度。由于数字化校园存在多个消费场景,使用Markov对策与Bayesian网络可以建立各个智能体之间的互相关系,从而增强各个智能体之间的交互性。由数据测试分析结果可知,文中所述方案识别贫困生的准确率可达80.9%,优于Adaboost算法。同时具有更低的平均绝对误差,适用于高校贫困生的资格认定。

It is difficult to identify the false application materials and accurately understand the real economic level of students in the traditional methods.In this paper,an accurate data analysis method based on the distributed reinforcement learning algorithm is proposed.This method collects the consumption data of middle school students in the digital campus,finds out the variables with classification ability after data preprocessing,and then improves the clustering speed by normalizing and selecting the appropriate number of hidden layers and neurons.Because there are many consumption scenarios in digital campus,using Markov game and Bayesian network can establish the relationship between agents,thus enhancing the interaction between agents.From the results of data test and analysis,we can see that the accuracy rate of the scheme can reach 80.9%,which is better than Adaboost algorithm,and has lower average absolute error,which is suitable for the qualification of poor college students.