Leveraging Uncertainty from Deep Learning for Trustworthy Materials Discovery Workflows
This work provides materials scientists with tools to improve the reliability and trustworthiness of machine learning models in materials discovery, particularly when dealing with microstructure analysis from SEM images.
This paper explores the use of predictive uncertainty from deep neural networks to address common challenges in materials discovery workflows. It demonstrates how uncertainty can determine necessary training data size, guide decision referral for confusing samples, and detect out-of-distribution test samples, improving classification model performance and dependability.
In this paper, we leverage predictive uncertainty of deep neural networks to answer challenging questions material scientists usually encounter in machine learning based materials applications workflows. First, we show that by leveraging predictive uncertainty, a user can determine the required training data set size necessary to achieve a certain classification accuracy. Next, we propose uncertainty guided decision referral to detect and refrain from making decisions on confusing samples. Finally, we show that predictive uncertainty can also be used to detect out-of-distribution test samples. We find that this scheme is accurate enough to detect a wide range of real-world shifts in data, e.g., changes in the image acquisition conditions or changes in the synthesis conditions. Using microstructure information from scanning electron microscope (SEM) images as an example use case, we show that leveraging uncertainty-aware deep learning can significantly improve the performance and dependability of classification models.