Abstract

In recent years, the Internet of Things (IoT) has developed rapidly due to its broad application prospects, and the types of sensing devices are becoming more and more abundant. In many applications, multi-dimensional attributes of monitoring objects are measured by deploying multiple independent heterogeneous data sources, thus heterogeneous multi-source multi-modal sensing data can be obtained. In this paper, we measure the data quality of multi-source and multi-modal data and make full use of the data quality information, a classification and detection method for heterogeneous multi-source and multi-modal sensing data was proposed. The purpose is to ensure the data quality and select some data sources for data transmission in order to save network resources as much as possible. Firstly, the problem caused by heterogeneous multi-modal data is solved by the feature transformation of deep neural network, and an innovative training method is used to extract the multi-modal shared feature expression with strong discriminant ability and low-dimensional characteristics from the original multi-modal high-dimensional data. Then, an algorithm combining deep learning with structural sparse multi-modal feature representation and mode selection is proposed, which innovatively uses deep learning to transform multi-modal data into modal-independent Abstract expressions. Finally, the structural sparse method is used to further select the feature dimension in the abstract expression to reduce the dimension of the final feature. Experiments show that the average false alarm rate of the proposed model is 2.2%, which is obviously superior to other similar algorithms.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call