What Makes a Successful AI Project? 5 Questions to Ask Before Training Your Neural Network

Speaker: Stephan Rasp, ClimateAI and TU Munich

The number of machine learning studies in weather and climate research is skyrocketing. This is only logical given the successes of AI in other domains. However, not every problem we face is necessarily suitable for a machine learning approach. In this talk, I will outline 5 key questions to ask before starting an AI project, which will help in designing studies that really progress the field, with examples from recently literature. Further, I will highlight a few specific problem areas, for example machine learning parameterizations, which pose additional challenges.