Reliable uncertainty estimation for seismic interpretation with prediction switches

Abstract

This paper presents a discussion on data selection for deep learning in the field of seismic interpretation. In order to achieve a robust generalization to the target volume, it is crucial to identify the specific samples are the most informative to the training process. The selection of the training set from a target volume is a critical factor in determining the effectiveness of the deep learning algorithm for interpreting seismic volumes. This paper proposes the inclusion of interpretation disagreement as a valuable and intuitive factor in the process of selecting training sets. The development of a novel data selection framework is inspired by established practices in seismic interpretation. The framework we have developed utilizes representation shifts to effectively model interpretation disagreement within neural networks. Additionally, it incorporates the disagreement measure to enhance attention towards geologically interesting regions throughout the data selection workflow. By combining this approach with active learning, a well-known machine learning paradigm for data selection, we arrive at a comprehensive and innovative framework for training set selection in seismic interpretation. In addition, we offer a specific implementation of our proposed framework, which we have named ATLAS. This implementation serves as a means for data selection. In this study, we present the results of our comprehensive experiments, which clearly indicate that ATLAS consistently surpasses traditional active learning frameworks in the field of seismic interpretation. Our findings reveal that ATLAS achieves improvements of up to 12% in mean intersection-over-union.

Contrastive Explanations In Neural Networks

“…Since there is no ground truth, uncertainty evaluation strategies utilize a set of samples rather than individual predictions for a pseudo-classification task. In [58], the authors utilize their proposed uncertainty metric to classify between in-distribution and out-of-distribution data. Higher the classification accuracy, better the uncertainty metric.…”

“…This research can be summarized by considering the sources of uncertainties, namely the data and model [27] uncertainties. A secondary research direction in UQ applies estimated uncertainty to select additional data for training [32], interpreting existing results [33], estimating image quality [34], and detecting out-of-distribution and adversarial samples [35]. In all cases, the statistical uncertainty of the decision is quantified.…”