Research On Enterprise Financial Risk Early Warning Based On Deep Learning Network

With the continuous development of China’s economy, the business environment are intricate,and the impact of foreign capital, the listing Corporation’s financial risk needs to be concerned. The accumulation of financial risk, will eventually form the financial crisis, this is a gradual process, so the financial crisis can be early warning. At present, there are a lot of methods used in the field of financial risk warning, such as multiple linear model,probabilistic model and BP neural network method. But these methods have some disadvantages, so we hope the method of deep learning applied to financial risk warning, to get better prediction effect.Deep learning of the application is currently booming, the combination of theory and practice, continue to promote the its development, and achieved remarkable results in the practical applications, such as Google’s Alpha go is the use of in-Deep Learning method to construct, in image recognition, speech recognition also made good results. Because, the deep learning has the self- learning effect, can process from the new data to study the characteristic unceasingly, therefore is one kind of adaptation very high method, this is the traditional financial risk early warning method does not have.In this paper, we first set up a financial risk early warning index system, includes seven categories of indicators reflect the characteristics of the enterprise. Down two 23 level indicators, in order to fully reflect the status of the enterprise. As the deep learning network needs to be trained in order to get the network parameters can be used for financial risk early warning. Therefore, this paper selects A-share non ST Listed Companies in 2014 data, and selected from 2008 to 2016 was ST Companies in the first two years of data as the training sample, which does not include the data of test samples, test sample from 2016 and by 2015,special treatment of the former two-year sample selection, which normal business from 2014 non ST companies selected. After training can be used for case analysis of deep learning network, select st Bo Yuan as the case enterprise, learning to network data of St Bo Yuan two years ago warning with depth of income. The results showed that with the financial crisis,and of St Bo Yuan failure are summarized. From the results of the analysis, deep learning method is applied to the financial risk early warning can achieve good results.Finally, the proposed method is put forward in the specific application of the specific application.