Improved Compact Inversion of Gravity and Magnetic Data using Huber Loss Function

This presentation explores the improved compact inversion of gravity and magnetic data using the Huber loss function to handle outliers and large residuals effectively. The approach combines robust statistical methods with geophysical inversion techniques to enhance subsurface imaging and interpretation.

Key highlights from the presentation:

• Demonstration of the Huber loss function for balancing sensitivity to outliers while preserving smooth model fitting.
• Application of regularization techniques to improve the stability and compactness of the inversion.
• Implementation of gradient-based optimization for resolving geological features with sharp contrasts.
• Case studies involving synthetic data tests and field datasets for validating the method’s performance.
• Comparisons with conventional least-squares inversion to showcase improvements in accuracy and robustness.

The results illustrate the potential of Huber loss-based inversion in scenarios where high noise or geological complexity can lead to model instability, providing a reliable tool for mineral exploration and subsurface characterization.