【摘要】 BP网络是一种典型的多层前向网络,由输入层、隐含层和输出层组成,通过学习样本训练模型后即可用于数据的预测,适用于实现网络教学系统的成绩预测功能。训练样本作为BP网络的学习数据集,对于BP网络模型的训练具有重要的作用。模型选取网络教学系统中能够影响学习成绩的相关因素作为输入数据,包括学生在线学习时间、学生学习能力、作业成绩和测试成绩,利用已有学生成绩作为训练BP网络的期望输出。将这些数据进行归一化处理即可用于训练BP网络模型。训练过程中,全局误差基本呈下降趋势,收敛效果较好。经过训练后的模型可预测出学生的成绩,并转化为相应的等级,对学生下一步学习进行指导,提出适合的教学策略。通过测试表明该模型可以用于教学系统中的学习成绩预测,获得了预期效果。更多还原

【Abstract】 BP network is a typical multilayer feed-forward neural network.It consists of three layers: the input layer,the hidden layer,and the output layer.After training by learning samples,the model can be used for data forecast,and is suitable to achieve the grade prediction of the network’s learning system.It is very import to select the training sample as BP neural network’s learning data set,informations can be got from students’ learning state tracking of the learning system,including online learning time,learning ability,homework score,and test score.Existing student grades can be used as expected output of the training BP network.These data need to be normalized.The globle error has a decreasing tendency and a good convergent performance during the training process.The grades can be forecasted from the trained model,then it provides a learning policy for strudent’s learning guide.The validations of the model show that the model can be applied to grade prediction,which has obtained expected results.更多还原