In underexplored regions, where well control is minimal, seismic facies analysis plays a crucial role in extracting geologic information. Furthermore, seismic attributes and machine learning techniques facilitate recognizing seismic facies in a quantitative and time-saving manner. We identify, primarily based on reflection patterns, several seismic facies in the Middle Miocene carbonate of the frontier Dangerous Grounds region. We employ texture attributes and an unsupervised artificial neural network to recognize those facies' natural grouping in the seismic volume. The basinal carbonates platform shows high amplitude, parallel, and highly continuous reflectors, whereas the marginal platform exhibits low to moderate amplitude, mounded reflection. The reefal buildups show mounded, divergent, and reflection-free patterns of weak internal reflectivity. The multiattribute analysis through the artificial neural network helps to quantify the targeted patterns, which is essential in understanding the depositional and diagenetic history of the carbonate buildups in the area. Copyright© (2022) by the European Association of Geoscientists & Engineers (EAGE). All rights reserved.