Unsupervised seismic facies classification using deep convolutional autoencoder

With the increased size and complexity of seismic surveys, manual labeling facies has become a significant challenge. Application automatic methods for interpretation could significantly reduce labor subjectivity particular interpreter present in conventional methods. A recently emerged group techniques is based on deep neural networks. These approaches are data-driven require large labeled data sets network training. We have developed convolutional autoencoder unsupervised classification, which does not manually examples. The maps generated by clustering deep-feature vectors obtained from input data. Our method yields accurate results real provides them instantaneously, allows an to identify dominant features. proposed approach opens possibilities analyze geologic patterns time without human intervention.