MACHINE LEARNING
“ Retail is a prime example of this in action – the rise of autonomous ‘ just walk out ’ stores have made many brands consider stripping out their own hardware and starting over to remain competitive , but it doesn ’ t have to be this way . Edge computing can be built into existing hardware – such as self-checkouts – and trained in-store . The result provides intelligent identification of non-barcoded items – such as fruit and veg – without investing in a whole new suite of tools ,” said Ben-Porat .
The Edgify CEO says that training these edge devices only takes a few hours , and the AI will continue to learn with each customer that uses it .
“ Every ‘ unknown item in the bagging area ’ is no longer an inconvenience for the store , but an opportunity for the machine to learn and become even more accurate . For incumbent retailers keen to provide a more modern shopping experience for customers , this approach will give a competitive advantage that doesn ’ t require the deep pockets of major tech brands such as Amazon ” he said .
Wider applications of edge technology This is only the beginning of worldwide rollout of TinyML across industries , according to Ben-Porat .
“ The impact of this technology extends
“ Edge networks help with the surge in performing AI tasks at the edge ”
GABRIEL AGUIAR NOURY PRODUCT MANAGER , CANONICAL
beyond commerce and has the potential to improve integral services such as healthcare . Patient data can be invaluable in giving medical professionals trends and insights on diagnosis and treatment . However , the need for this information to remain confidential is a major hurdle to overcome . Traditional cloud solutions have automatically been discounted , due to the inherent security issues present in moving data outside the hospital ,” he said .
Again , Ben-Porat notes that the benefits lie in using existing technology and upgrading it . He added : “ Embedding edge devices within medical tools such as MRI machines can be a more effective solution . Medical professionals can gather this data , identify trends and still share valuable insights – all without compromising the privacy of the patient . The result will improve processes , treatment and the overall patient experience .”
technologymagazine . com 127