Deep learning algorithms in enterprise accounting management analysis

This paper first constructs two accounting management prediction models. The first one is the CNN-GRU model, in which samples are input to the CNN model for extracting high-level abstract features by convolution and pooling and then input to the GRU model to train time-series potential features. The other is the CNN+GRU fusion model, where both CNN and GRU models are Merge fused, and the abstract features extracted from the two models trained separately are stitched together and then trained together. Then, through experimental comparison with other machine and deep learning methods, the two models developed in this study er on every evaluation index. The CNN+GRU Finally, the index system of accounting management is established, including seven categories of indicators reflecting the characteristics of the enterprise, which are divided into 23 secondary indicators, and the prediction of the enterprise accounting management of ST Boyuan in 2013 and 2014 is made by using the obtained deep learning. The results show that: the judgment rate of the enterprises in the year before the crisis is higher than that of the enterprises in the two years before the crisis, but both reach more than 90%, and the overall prediction accuracy rate reaches 95%, indicating that using deep learning for accounting management prediction can achieve very good results. The research results of this study help the enterprises’ internal management and thus have a guiding influence on their economic development.