Sensing Urban with Wi-Fi and Satellite

Chen Yan,Peng Wang,Lifeng Sun
DOI: https://doi.org/10.1145/3126686.3126721
2017-01-01
Abstract:Modern urbanization is featured with rapid development and collective intelligence, for which how to effectively utilize multimodal urban sensory data is a hot topic. In this paper, we focus on automatically recognizing functional urban regions by fusing two important sensory media, including collective behavioral information reflected by Wi-Fi accessing records collected from large populations as well as high-resolution satellite urban imaging. An aggregation algorithm is firstly applied to identify candidate functional regions characterized by geographical clusters of Wi-Fi access points. Semantic behavioral and visual features are extracted from two heterogeneous sensory data respectively, from which multi-functional properties are recognized by combining decision trees with boosting method. Finally, a transfer learning algorithm derived from AdaBoost is proposed to adapt the trained model to other cities with different feature distribution. Experimental results on real data sets demonstrate the efficacy of the proposed model in terms of recognition accuracy.
What problem does this paper attempt to address?