An Image-Driven Machine Learning Method for Microstructure Characterization in Metal Additive Manufacturing: Generative Adversarial Network

Abstract The recent development of artificial intelligence especially machine learning technology has provided an emerging direction for solving microstructure representation and analysis in additive manufacturing. In this work, we introduce an advanced image-driven machine learning algorithm that offers an effective way to abstract the features in microstructure and generates high-resolution and large-size images that can represent the original counterparts. The evolution of the model and the potential application of the algorithm in material science are also discussed.