Application of Support Vector Machines in Inverse Problems in Ocean Color Remote Sensing

Abstract

Neural networks are widely used as transfer functions in inverse problems in remote sensing. However, this method still suffers from some problems such as the danger of over-fitting and may easily be trapped in a local minimum. This paper investigates the possibility of using a new universal approximator, support vector machine (SVM), as the nonlinear transfer function in inverse problem in ocean color remote sensing. A field data set is used to evaluate the performance of the proposed approach. Experimental results show that the SVM performs as well as the optimal multi-layer perceptron (MLP) and can be a promising alternative to the conventional MLPs for the retrieval of oceanic chlorophyll concentration from marine reflectance.