New deep learning system design technique could boost edge image analysis
A new technique for boosting the performance of deep learning systems with possible implications for computer vision applications has been developed by DarwinAI co-founder, Canada Research Chair in AI, and University of Waterloo and Professor of Systems Design Engineering Alexander Wong. The system is based on a compact family of neural networks that could run on embedded and mobile devices such as smartphones and tablets, according to an announcement by the University.
Wong calls the neural networks “AttoNets.” They can be used for segmentation and classification computer vision tasks, and be used to build action recognition, image generation, and other visual applications. Human designers work with the AI to design new networks, which enables high-performance results on edge devices, according to the announcement.
“The problem with current neural networks is they are being built by hand and incredibly large and complex and difficult to run in any real-world situation,” says Wong. “These on-the-edge networks are small and agile and could have huge implications for the automotive, aerospace, agriculture, finance, and consumer electronics sectors.”
DarwinAI was founded to commercialize this technology.
The approach is called Generative Synthesis, and is was recently validated by Intel, and has shown promise in research with Audi Electronic Ventures. The company has also argued it can improve algorithmic transparency.
“We took a collaborative design approach that leveraged human ingenuity and experience with the meticulousness and speed of AI because a computer can crunch really fast,” Wong explains. “It’s already having a real-world impact, especially where there is a need for these on-the-edge deep learning solutions to power infrastructure and intelligence systems or protect user privacy.”