Machine Learning Approaches for Multi-Sensor Data Pattern Recognition: K-means, Deep Neural Networks, and Multi-layer K-means

摘要

Learning data patterns from spatially distributed sensors is an important step towards accurate situational awareness in applications such as traffic management, surveillance, and border security. One key problem in learning data patterns is to uncover the data-object association for the prediction and estimation of object distribution. In this paper, we focus on presenting three machine learning approaches, namely, K-means, deep neural networks, and the newly developed multi-layer K-means (MLKM). The MLKM is based on leveraging the advantages of some existing machine learning approaches, including K-means, K-means+ + , and deep neural networks. To enable the accurate data association from different sensors for efficient object localization. MLKM relies on the clustering capabilities of K-means+ + structured in a multi-layer framework with the error correction feature that is motivated by the backpropogation that is well-known in deep learning research. To show the effectiveness of the MLKM method, numerous simulation examples are conducted to compare its performance with K-means, K-means+ + , and deep neural networks.