The AI That Aims to Break Free From Wi-Fi Dependence

A new form of artificial intelligence (AI) that doesn’t need internet access could become a powerful healthcare tool, according to one of its creators.

Researchers and technicians from the University of Waterloo in Ontario, Canada recently presented the potentially groundbreaking new deep learning AI technology at the International Conference for Computer Vision in Venice, Italy. The team made the AI less reliant on iCloud and Wi-Fi systems, using deep intelligence that mimics nature by incorporating “sexual evolutionary synthesis” to improve deep neural networks, according to the study.

This new technology has the potential to improve communication in a wide range of industries, including healthcare, where such AI could give physicians and healthcare staff the edge in crisis situations, a study author told Healthcare Analytics News™.

“This technology, which produces highly efficient and compact deep learning AI that can run on low-power devices without being tethered to the internet, can be a great enabler for producing AI software that can aid clinical practitioners to make more accurate and faster clinical decisions and treatment planning directly from their smartphones, tablets, and other handheld devices from wherever they are,” says co-inventor Alexander Wong, PEng.

He is the Canada research chair in medical imaging systems and co-director of the Vision and Image Processing Research Group in the University of Waterloo’s Department of Systems Design Engineering. He and his teammates came up with the name “evolutionary synthesis” because the AI “attempts to mimic processes in biological evolution within a probabilistic framework to synthesize very compact deep neural networks,” Wong says.

The results: A network synthesized in this manner had “double the architectural efficiency” of a network touched by asexual evolutionary synthesis, according to the study. Both networks boasted 97% testing accuracy.

Sexual evolutionary synthesis can be useful in developing countries where clinical specialists are less widely available, Wong explains. Many people in underdeveloped countries have access to low-cost smartphones on which the AI can be installed, enhancing public health, for instance, through water quality testing. “Water quality is a big issue in underdeveloped countries, as safe drinking water is scarce, and this can be used to help them test their water more readily,” Wong says.

Across the world, this brand of efficient yet powerful AI can augment the healthcare system to provide better quality of care, he adds.

Surgeons who guide robotic operations could receive assistance from sexual evolutionary synthesis. “This new AI technology may be leveraged to build more efficient and compact deep learning AI that can sit directly within the surgical robots, where reducing lag and latency is very important and relying on the internet is not well-suited for such real-time, mission-critical operations,” Wong says. During their trials, researchers installed the technology in small chips in smartphones and tablets, opening the door to additional possibilities.

In the last decade, hospitals have struggled through power outages due to hurricanes and flooding. “Because this technology can build more efficient and compact deep learning AI, the energy requirements for running that AI can be lower,” Wong says, “and so [it] is more suitable for cases during a blackout where only a more limited amount of backup power or batteries are available.” It also enables deep learning AI to run on mobile devices, allowing low-power devices to be widely used during crises, he notes.

The software is entering the commercial realm this year through a start-up called DarwinAI, which Wong co-founded.