Insights into Nanodiamond from Machine Learning

Nanodiamond is a complex material which is difficult to simulateSimulate, synthesiseSynthesise and characterise. Data sets tend to be small but a range of machine leaning methods are available to leverage the information contained in collection of studies that has been published over the past 20 years. In this chapter a series machine learning studies of nanodiamond are surveyed, showing how different approaches can extract new knowledge from a single data set. This includes unsupervised learning for dimension reductionDimension reduction and clusteringClustering to find hidden patterns based on intrinsic similarity; supervised learningSupervised learning to predict unique classes or structure/property relationships; inverse design to identify which structure to make (or separate) and statistical learningStatistical learning to unlock the underlying causes. The value of evaluation methods including explainable AIArtificial Intelligence (AI) are demonstrated, and new directions based on advanced machine learning methods are purposed. With this chapter readers will gain an overview of types of insights that can be obtained from machine learning, and how they may be applied to other diamond data sets in the future.