Artificial intelligence (AI), one of the most important aspects of our future digital age, is currently experiencing a major boom in science, the economy and the media and has already become an everyday technology. Thanks to SIRI we can now speak to our smartphones, the first self-driving cars have started to appear on the roads of California and logistics providers are already testing autonomous flying drones. However, if machines are to be used safely in factories, hospitals and households they must – just like us humans – be able to act, react and learn by observation and experience and not simply operate in accordance with pre-programmed data.
Hybrid research methods:
Combining knowledge-driven and data-driven approaches to close the »semantic gap«
»Multimedia Pattern Recognition« and »Data Science« are the Fraunhofer IAIS’s core areas of research and most important areas of expertise in relation to artificial intelligence and its applications in robotics, image and speech processing and process optimization. Central to the research being done at the Fraunhofer IAIS are hybrid artificial intelligence solutions: This is where we combine the knowledge based research methods of professors Dr. Sören Auer and Dr. Jens Lehmann with the data-driven methods by professors Dr. Stefan Wrobel and Dr. Christian Bauckhage.
The knowledge-based approach is founded on people’s predetermined empirical knowledge and the conclusions drawn from it – and has recently been encapsulated in the phrase »Semantic Web«. The data-based approach on the other hand uses »Machine Learning« methods to analyze statistical correlations. The aim of combining both approaches is to close the »semantic gap«. This is where intuitive empirical knowledge and statistical knowledge meet and need to be put into context if they are to replicate the human ability to understand meaning from a given context.