- Team
- 2 people
- Duration
- 40 months
- Components
- AI tool for Earth system data with reproducibility and collaboration on HPC infrastructure
- Models
- HPC connectors, big data, MLOps, AI
- Language
- Python, Typescript
- Tools
- HPC, Serverless, Docker
KI:STE - AI Strategy for Earth system data
Objectives and solutions
The KI:STE project aims to exploit recent developments in artificial intelligence – especially deep learning methods – for sound environmental data analysis. The scientific goal is the implementation of current AI approaches for spatiotemporal variable pattern recognition and pattern analysis in environmental data in the subject areas clouds, snow/ice, water, air quality and vegetation within the framework of five dissertations. In addition, a technical platform will be created to make powerful AI applications on environmental data available in a portable way. An online AI learning platform will be established with Interfaces to this AI platform. This e-learning offer is aimed at the location-independent education of young scientists and other interested parties. It will use the concepts and methods developed in the five research fields as teaching material.
In KI:STE, we address topics, which are not particularly easy: data are big, complexity is high and the context (weather and climate) is a whole scientific field! With respect to machine learning in this scenario, one of the most prominent questions concerns the optimization of models with respect to size, reliability under variation of data, and for certain hardware. Obviously, this question is important for any AI problem, consequently, we generalize the issue and invite you to participate in our online experiment.
→ Questions or thoughts about AI in environmental analysis? Dr. Markus Abel and Thomas Seidler are looking forward to your message.