Making AI models up to 90% more energy efficient (IMAGE)
Caption
Powering artificial intelligence comes with a massive energy bill attached. Professor Wolfgang Maaß and his research team at Saarland University and the German Research Center for Artificial Intelligence (DFKI) want to make AI up to 90% percent more energy efficient. To improve AI’s carbon footprint and to reduce costs, the Saarbrücken team is rethinking data centres, large language models and image analysis models – and their research is opening up access to powerful AI models for small and medium-sized companies. One of the methods that researchers Sabine Janzen (right) and Hannah Stein (left) are using is known as ‘knowledge distillation’ – a form of model compression that transfers knowledge from complex models to simpler ones. After all, we don't need to read an entire library to answer a specific question. Instead, we focus on those books that are relevant to the question at hand. From 31 March to 4 April, the researchers will be showcasing their work at this year’s Hannover Messe at the stand of the Federal Ministry for Economic Affairs and Climate Action (Hall 2, Stand A18).
Credit
Credit: Oliver Dietze
Usage Restrictions
Press photographs can be used free of charge with this press release or in connection with reports about Saarland University provided that a photo credit with the photographer’s name is included.
License
Original content