Machine learning models for energetic materials properties using multi-task learning

Data science and artificial intelligence are playing an increasingly important role in the physical sciences. However, many fields, including energetic materials, suffer from scarce data, and the available data is not organized in a way conducive to machine learning. To address this gap, we identified rich sources of experimental and calculated data and collected this data into an electronic format that is tailored for efficient querying, filtering, and extracting. We will present new predictive models that use multi-task learning to learn multiple properties and address data scarcity.