Research on Deep Learning Feature Extraction Algorithm for Big Data Analysis

In this study, a novel Spatio-Temporal Convolutional Recurrent Network (ST-CRN) is proposed, this methodology integrates the respective advantages of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), specifically designed to handle complex datasets that involve both spatial and temporal dimensions. CNNS are good at extracting spatial features of data with grid structure, while RNNS are particularly suitable for processing sequence data, but they each have limitations in processing high-dimensional spatial data or time series data. By integrating the CNN's spatial feature extraction capabilities with the RNN's time series modeling prowess, ST-CRN notably enhances the comprehension and classification performance for intricate spatiotemporal data. Experimental outcomes demonstrate that ST-CRN exhibits considerable superiority over both CNN and RNN in terms of key assessment metrics, including accuracy, recall rate, and F1 score. While ST-CRN's computational time is marginally longer than that of CNN, it still maintains an efficiency edge compared to RNN. In summary, ST-CRN not only improves the performance of complex spatiotemporal data analysis tasks, but also achieves a good balance between accuracy, completeness and computational efficiency, becoming an ideal choice for processing multidimensional complex data, and providing a new direction for future research.