The goal of financial accounting is to provide users with quantitative financial information about enterprises, thereby helping users make financial decisions. Financial information providers should not only consider how to provide information to users, but also how to provide information to users in the best and fastest way. In financial information processing, it is necessary to fully consider users’ expectations of financial information, and establish a matching relationship with users’ input and output interfaces of financial information, thereby improving the efficiency of using financial information. This paper discussed the financial information automatic processing system that improves SDAE (Stacked Denoising Autoencoder) algorithm, and systematically introduced the problems of improving SDAE network algorithm, financial information automatic processing enterprise operation and development, and the technical characteristics of big data such as data processing and data mining. Finally, the superiority of the financial information automatic processing system based on the improved SDAE network algorithm was verified through experiments (the reconstruction error of the first group of improved algorithms decreased by an average of 1.6%).