Researchers at UWM are looking for ways to reduce the amount of data that is processed at data centers while still getting good computing results and saving energy.
Shuaiqi Roger Shen, assistant professor of electrical engineering, said this research, powered by a tiny GPU processor, also called an edge-of-network AI device. He pairs efficient algorithms with edge processing by these small devices right at the source of the data. This kind of processing keeps the data private and out of the cloud and out of data centers.
It’s one strategy that improves efficiency both for data centers and industries using AI, like manufacturing. Shen’s doctoral student, Monika Gawande also was interviewed. She said the device works like a mini data center, enabling AI to comb through a lot of information very quickly.
She described how the GPU technology allows her to “prune” the processing — reducing the data that needs to be stored and keeping only the essential information required by AI to do the job. Watch the report.