Аннотация:The inverse problem (IP) of electrical prospecting is a complicated high-dimensional ill-posed problem with a well-known instability. To describe the sought distribution of the electrical conductivity, different parameterization schemes are used. The most general scheme uses the values of conductivity in the nodes of a pre-defined grid, with further interpolation between nodes. More specific schemes may assume presence of certain geological structures. Transfer from the solution of the IP within general scheme to its much more stable solution within one of specific schemes in a narrower class of geoelectric sections causes the necessity of prior classification of the studied data pattern, resulting in the selection of the most appropriate parameterization scheme. In their previous studies, the authors considered the solution of the IP of magnetotelluric sounding (MTS) using artificial neural networks (ANN) (perceptrons). Also, it was demonstrated that the described classification problem can be successfully solved by ANN with average rate of correct determination of the parameterization scheme exceeding 97%. Since then, the authors have elaborated a novel method of ANN-based solution of the MTS IP within scheme G0, based on simultaneous determination of a group of several parameters at once. In this study, the developed method has been extended to other parameterization schemes. It is demonstrated that group determination of parameters is an effective method for ANN solution of the MTS IP for any parameterization scheme. In future studies, it is planned to test the complex algorithm combining classification and IP solution within partial classes of geoelectrical sections against the general approach within the most general parameterization scheme G0.