Enhanced Behavior of Adaptive Neuro-Fuzzy Inference System (ANFIS) Algorithm over Artificial Neural Networks (ANN) in Application of Geoelectrical Resistivity Data

Soft computing techniques are widely used for many non- linear problems in the real world. Many Earth's nonlinear characteristics exhibit uncertainty problems that have to be interpreted with advanced soft computing tools. Ambiguity always presents in realistic processes. The efficiency of knowledge–based systems depends upon the algorithms, which are cumbersome as their implementations require extensive computational time. Here, we present a work about interpreting the subsurface parameters of the Earth from electrical resistivity data using the Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy inference (ANFIS) techniques. We focus on the advantage of the hybrid neuro-fuzzy systems, compared with the Artificial Neural Networks (ANN), in efficiency in interpreting electrical resistivity data. Hybrid systems that fuse fuzzy systems and neural networks (NN) have been propounded for utilizing numerical data. It expected that ANFIS can be used in many nonlinear problems. The network model is successful in training with large number of self-generated synthetic data sets. The interpretation using the ANFIS technique gave promising results with better accuracy, compared with the ANN inversion. Problems with parameter estimation can be solved more efficiently with this ANFIS geoelectrical resistivity inversion algorithm.