Neural Networks and Support Vector Machines and Their Application to Aerosol and Cloud Remote Sensing: A Review

Abstract

Machine learning techniques, such as artificial neural networks and support vector machines , are becoming increasingly popular in the remote sensing community. They can be used to solve inverse problems as well as for data classification and clustering. The first applications of machine learning methods to remote sensing problems were mainly aimed at tasks such as land use classification , identification of specific objects (e.g. clouds) in satellite imagery, and atmospheric profiling. In the last decade, these methods have started to receive attention in the aerosol and cloud remote sensing community as tools to speed up the retrieval of aerosol and cloud properties. Machine learning methods can enter the processing chain of a remote sensing product in several ways. They have been used as stand-alone retrieval or classification algorithms, as fast approximate forward models or as part of a more complex type of algorithm. In this paper we review examples of use of machine learning techniques in the three ways mentioned above. Furthermore, we discuss the theoretical basis underlying the use of these techniques in remote sensing , as well as their advantages and disadvantages with respect to the traditional processing schemes.