With the rapid development of society today, in the context of the impact of education on personal future development, people have increasingly attached importance to education. Various types of online and offline education institutions have mushroomed. However, the factors that affect the effectiveness of education are numerous and complex. As a very important part of early childhood learning, preschool education should receive more attention. The deep learning algorithm can reasonably classify and plan the factors affecting preschool education through massive data calculations to ensure the reasonable allocation of resources and save the time and energy costs wasted in allocating educational resources. The educational effectiveness of preschool education institutions and organizations can be reasonably evaluated. This article applied deep learning algorithms to the evaluation of preschool education effectiveness. Data related to the effectiveness of preschool education in preschool education institutions using convolutional neural networks and not using algorithms were compared. The experimental results showed that the expert recognition rates of the convolutional neural network algorithm and traditional manual measurement data were 98.8% and 92.55%, respectively. At the time level of teaching quality estimation, five algorithms and 100 sets of relevant experimental data were compared. In terms of the accuracy of feature search, the average accuracy of the convolutional neural network algorithm was 97.96%, while the accuracy of the traditional manual search was 53.9%. Therefore, the application of the convolutional neural network algorithm in deep learning to the evaluation of preschool education effectiveness was more efficient and efficient and could improve the evaluation effect from various aspects.

当今社会飞速发展，在教育影响个人未来发展的背景下，人们对教育越来越重视。各类线上线下教育机构如雨后春笋般涌现。然而，影响教育效果的因素众多且复杂。学前教育作为幼儿学习的一个非常重要的组成部分，应该受到更多的重视。深度学习算法可以通过海量数据计算，对影响学前教育的因素进行合理分类和规划，保证资源的合理配置，节省分配教育资源所浪费的时间和精力成本。学前教育机构和组织的教育效果能够得到合理评价。本文将深度学习算法应用于学前教育效果评估。对使用卷积神经网络和不使用算法的学前教育机构学前教育效果相关的数据进行了比较。实验结果表明，卷积神经网络算法与传统人工测量数据的专家识别率分别为98.8%和92.55%。在教学质量评价的时间层面上，对5种算法和100组相关实验数据进行了比较。在特征搜索的准确率方面，卷积神经网络算法的平均准确率为97.96%，而传统人工搜索的准确率为53.9%。因此，将深度学习中的卷积神经网络算法应用于学前教育效果评估更加高效快捷，可以从多方面提高评估效果。