Performance Optimization Design of Big Data Processing System Based on Deep Learning Algorithm

With the rapid growth of information technology, the data in this field shows an explosive growth trend, which also means the arrival of the “big data” era. This article uses Deep Belief Network (DBN) to improve the big data processing system, shorten the life cycle of data processing, fully improve the efficiency of data processing, and provide a more convenient platform technology for data processing, so as to promote the growth of big data. The simulation results show that the accuracy of the big data feature selection model can reach more than $96 \%$, which has obvious advantages over the random forest algorithm. Deep learning (DL) algorithm extracts and classifies the complicated data in the data connected subgraph, thus extracting the key information needed by users. Effective management of computing resources through this algorithm can provide a lightweight virtualization environment for computing nodes with almost no extra overhead, and improve the management efficiency and resource utilization of computing platforms.