Precision in Energy-aware AI Research for Low-carbon Solutions
Artificial intelligence (AI) and machine learning (ML) have immense potential in a wide variety of areas, but require improved reliability and a reduced carbon footprint to make their use safer and bring them into line with measures to mitigate the climate crisis.
In our project PEARLS, we develop methods that:
- Estimate the uncertainty and resource consumption of AI models
- Test methods to reduce the size and complexity of AI models
- Investigate more native storage formats
- Apply our findings to motivate the reduction of the carbon footprint in scientific research in High Energy Physics (HEP) and other fields
Because every gramm of CO2 kept away from the atmosphere keeps our planet cooler and safer for us to live and do research.