Dates: 2024
Principal Investigator: Dr.Korhan Ayranci
Description: Geosciences, especially in petroleum geology, rely heavily on visual observations and multidisciplinary research for accurate results. Lithofacies analysis is crucial for understanding reservoir heterogeneity in oil and gas exploration, but it requires extensive training and can lead to errors. Leveraging deep learning techniques, particularly convolutional neural networks (CNNs), offers a promising solution to automate and improve lithofacies analysis. By investigating existing CNN architectures, optimal feature learning and classification of geological datasets can be achieved. However, ensuring the interpretability of the model's results is equally important. To address this, state-of-the-art explainable machine learning modules will be incorporated, providing insights into how the model arrives at its decisions. This interpretability feature is lacking in current geological machine-learning models. By combining deep learning for accurate analysis and explainable AI for transparent decision-making, this research aims to revolutionize lithofacies analysis, reducing errors and bias while enhancing efficiency for geoscientists.