Machine learning techniques have been applied to seismic interpretation to help identify seismic patterns, which are difficult to map, especially, in new discoveries and large volumes of seismic data. This work aims to apply a methodology for identification and characterization of carbonates facies in the Barra Velha Formation, on the Wildcat Prospect in the Santos Basin, from seismic attributes and a non-supervised facies classification. The machine learning method used is the Self-Growing Neural Network (SGNN). In our workflow, we performed the following steps: (i) carbonate seismic patterns identification through seismic amplitude, where was possible to identify the build-up facies, characterized by chaotic seismic textures with a conical external geometry and internal fracturing; debris facies, exhibit prograde geometry with chaotic internal texture; carbonate platform facies showing a well-defined flat parallel reflectors; and the bottom lake facies, that does not have specific geometry and internally the reflectors are chaotic; (ii) seismic attributes generation and analysis of seismic patterns, the chosen attributes were eigen coherence, dip steered enhancement, relief and relative acoustic impedance. (iii) Then, we performed a principal component analysis (PCA), with seismic amplitude filtered from dip steering enhancement (DSE), eigen coherence and relief as inputs, and (iv) Finally, we carried out the classification using the seismic attributes and PCA results. It was possible to differentiate between the carbonate platform from fractured areas, especially those related to the build-ups. This is important for field development, as in the study area, build-ups represent the best reservoirs. Even from the PCA, which helps in a better clustering, one of the difficulties encountered has been to differentiate the build-ups from the debris facies. To overcome this challenge, other seismic attributes are being evaluated. Finally, our results help in a better tectono-stratigraphic understanding of an important prospect in the Santos Basin, which is the most important basin in hydrocarbon production in Brazil, accounting for more than 70% of all production in the country.