Machine learning (loosely referred to as Artificial Intelligence) might sound futuristic, but you almost certainly experience it in your daily life, whether it's Netflix suggesting what to watch next, Siri responding to voice commands, or your bank automatically flagging a potentially fraudulent transaction.
One aspect of machine learning that is perhaps most applicable to our industry, is its ability to recognise faces, handwriting, and other images. This means it can scan vast amounts of data much faster, and often more reliably, than a human.
In a recent Microsoft experiment, over the course of a day, a team developed an application that scanned the entire public release of the JFK files totalling 34,000 pages and attributed a kind of meta data to each page [read more]. This information would have taken decades for a human to read, however users can now search by keyword (Oswald, for example) and return hundreds of images, handwritten documents, and typed information that relates to that keyword – search The JFK Files for 'Oswald' to try it for yourself.
Another example of machine learning is in roadside cameras and connected systems that optimise traffic flows, by predicting future events. Employing the same technology as driverless vehicles, machine learning code inside roadside cameras distinguishes between pedestrians, cyclists, motorcycles, cars and trucks, on the road network, transmitting refined traffic ‘count data’ back to the core system, in real time.
'Edge processing' (implemented by us in recent projects), means each camera unit can process its own video feed before transmitting it back to the central system using onboard mobile technology, this avoids transmission of bulky video containing personally identifiable information. Once these data reach the main system, they are analysed using similar machine learning techniques, enabling the system to accurately predict changes in traffic demands, an hour ahead of when events will occur. This drives next-generation traffic signal optimisation, to smooth traffic flows, and reduce congestion. Find out more at Vivacity Labs
Application programming interface's (API's) from the likes of the Met Office and Environment Agency, can also be used to enrich data from roadside traffic counters with behavioural weather and environmental-based trends to see how an imminent environmental event might affect changes in traffic flow. Indeed, the Smart London Plan talks about “how fixed and mobile sensors across the city, and intelligent connected vehicles, can be used in the collection of weather, emission and traffic flow data, for use by city planners in the development of more sustainable future cities.”
Despite these innovative applications of this technology, we're only beginning to scratch the surface of the real-world potential machine learning can offer.
How we can help
By leveraging the multi-billion-dollar investments into Microsoft's Azure platform and its machine learning components, our Microsoft Certified team members can apply immense cloud computing power to the software we are developing, enabling us to provide futuristic software capability to our clients, cost effectively. In collaboration with the wider infrastructure team, we are able to help realise a Smart Cities vision, using artificial intelligence to drive the software elements of multidisciplinary projects. Ask our consultants how they might help you, check out our other 2018 tech trends below.
Capita Real Estate and Infrastructure work with public and private sector organisations to design, build and optimise their real estate and infrastructure assets. From thought to finish, we apply our combined expertise to achieve more from the entire built environment.
We build competitive advantage through intelligently applied real estate and infrastructure solutions and enhance our clients’ standing in a forever-changing world.