AI Hardware Acceleration
The increasingly complex AI models require a lot of energy and power. Conventional computer architectures quickly reach their limits here. Therefore, engineers are increasingly working on the development of hardware specialized in AI. Thanks to the more efficient and faster calculation of AI models on special hardware, the increasingly complex applications can even be used on mobile devices. But to be able to use AI software and hardware, you need the necessary know-how. And that is what is missing. Studies show that the shortage of skilled workers prevents small and medium-sized companies in particular from using AI solutions and possibly threatens their competitiveness.
Help comes from the universities: The AI qualification project HAISEM-Lab will offer one- or two-day courses especially for employees of small and medium-sized enterprises from 2020. Project partners are the L3S, the Institute of Microelectronic Systems (both Leibniz Universität Hannover) and the Software Systems Engineering group of the University of Hildesheim. Together, they use their extensive experience to provide the participants not only with basic AI competencies but also with the necessary methods to develop and apply AI accelerators on different hardware platforms. For this purpose, the project partners provide the participants with special hardware such as GPUs, FPGAs and TPUs. Special emphasis is placed on the practical parts and the practical phases of the courses so that the participants can directly apply and deepen their acquired knowledge.
Sergej Zerr was a research group leader at L3S and deals with AI for intelligent production. He headed the HAISEM-Lab and the research laboratory “Society and Work”.