Into the Noddyverse: a massive data store of 3D geological models for machine learning and inversion applications

[摘要] Unlike some other well-known challenges such as facial recognition, wheremachine learning and inversion algorithms are widely developed, thegeosciences suffer from a lack of large, labelled data sets that can be usedto validate or train robust machine learning and inversion schemes. Publiclyavailable 3D geological models are far too restricted in both number and therange of geological scenarios to serve these purposes. With reference toinverting geophysical data this problem is further exacerbated as in mostcases real geophysical observations result from unknown 3D geology, andsynthetic test data sets are often not particularly geological orgeologically diverse. To overcome these limitations, we have used the Noddymodelling platform to generate 1 million models, which represent the firstpublicly accessible massive training set for 3D geology and resultinggravity and magnetic data sets ( https://doi.org/10.5281/zenodo.4589883 , Jessell, 2021). This model suitecan be used to train machine learning systems and to provide comprehensivetest suites for geophysical inversion. We describe the methodology forproducing the model suite and discuss the opportunities such a model suiteaffords, as well as its limitations, and how we can grow and access thisresource.

[学科分类] 眼科学