Algebraic reconstruction technique for neuro-fuzzy geotomography

A new algebraic reconstruction techniques (ART) are developed for a neuro-fuzzy geotomography to accelerate the convergence of learning phase and to reduce the learning time or iteration times. The learning algorithm is derived from a constrained optimization problem. The Minkowski norm of the corrections of parameters is used as the objective function of the optimization problem. Some computer simulation results show that smooth distributions of a material parameter are obtained by using the Minkowski norm. Furthermore, the proposed method is applied to the experimental data collected at a dam site by cross borehole seismic probing.