11.00 – 11.25
In industrial AI applications and projects, success depends not only on high-quality real data, but equally on the meaningful interpretation of that data at the edge. Not knowing the AI’s interpretation (black box trap) can quickly lead to wrong decisions by misinterpretation and malfunction. By combining real-world datasets with synthetic data for model training and analysis of AI decision processes, it is possible to close critical gaps. On the AI inference side, robustness and efficiency can be achieved with the latest AI acceleration architecture for Edge AI using the ‘open’ RISC-V technology.
This talk illustrates how to leverage both data sources effectively, avoid misinterpretation, and build reliable AI solutions for real-life industrial environments.
Speaker: Syamak Nazary – Axelera.AI (on behalf of RUTRONIK Elektronische Bauelemente GmbH)