The classification of urban functional zones is crucial for improving land use efficiency and promoting balanced development across urban areas. Existing methods for classifying urban functional zones using mobile signaling data face challenges primarily due to the limitations of single data sources, insufficient utilization of multidimensional data, and inherent inaccuracies in mobile signaling data. To address these issues, this study proposes an innovative classification method that employs advanced multimodal data fusion techniques to enhance the accuracy and reliability of functional zone classification. Mobile signaling data are mapped into image data using timestamp and geographic location information and combined with point of interest (POI) data to construct a comprehensive multimodal dataset. Deep learning techniques are then applied to fuse the multimodal data features, enabling precise and reliable classification of functional zones. The experimental results demonstrate that this method achieves an accuracy of 95.128% in classifying urban functional zones, significantly outperforming methods that use single-modal data.

Video action recognition faces persistent challenges in balancing accuracy with computational efficiency. While state space models, such as Mamba, have emerged with linear complexity advantages, they exhibit inefficiency in capturing critical spatiotemporal dependencies within video data. To address this core limitation, this paper proposes ETMamba, an enhanced architecture built upon the Mamba baseline. The ETMamba achieve performance breakthroughs via three core innovation modules: (1) the Spatiotemporal Feature Preservation module retains complete original spatiotemporal correlations before data flattening, solving the problem of spatiotemporal feature loss; (2) the Efficient Bidirectional Sharing strategy accurately models bidirectional temporal dependencies, enhancing key temporal dynamic information; and (3) the Spatiotemporal Collaborative Modulation mechanism combines global temporal and local spatial information to achieve collaborative capture of long-short term dependencies and fine-grained features. We conduct experiments on multiple benchmark datasets, achieving recognition accuracies of 88.3%, 74.6%, 75.7%, and 98.1% on Kinetics-400, Something-Something V2, HMDB-51, and Breakfast datasets, respectively, while maintaining low to medium computational complexity.