Deloitte: Machine learning will reach 'prime time' status by 2020

Jan 9, 2018 | 5739 Views

A lower barrier to entry and a spike in the number of pilot projects involving machine learning will lead to various industries doubling its implementation this year and again in 2020, according to Deloitte Canada.

"Machine learning technology is going prime time by 2020," Duncan Stewart, Deloitte Canada's director of technology, media and telecommunications research told ITbusiness.ca before the company's 2018 Technology, Media & Telecommunications (TMT) Predictions roadshow in Toronto. "Between 2009 and 2017, there were some challenges when it came to wide-scale adoption. But there's going to be an elimination of barriers in 2018."

Steward points to five specific factors that will make machine learning more accessible: automation, data reduction, chip upgrades, added transparency and local machine learning.

There are a number of tools that automate the collection and preparation of data for machine learning and it will open the doors for companies that don't have in-house teams of scientists and researchers to adopt the technology, he says.

Data reduction is another, allowing smaller companies to tap into the technology without having to rely on enormous amounts of data points larger companies such as Bell and Rogers have access to.

"That, of course, expands the market and lowers the price of machine learning," he says.

       

Field programmable gate array (FPGA) and application specific integrated circuit (ASIC) chips will also learn new tricks, says Stewart.

"These chips have been around for years but the use for accelerated machine learning is new. The two tend to be cheaper and use significantly less power than GPUs, so for certain applications they are capable of, literally, according to people involved in machine learning, to do the same workload at one-tenth the power consumption the previous technology did," says Stewart.

Innovations in local machine learning, which allows for machine learning to occur on personal devices without having to communicate over a network, he says, will also increase the technology's adoption in several industries.

Unlocking the black-box

It's difficult for people, even behind the wheel of advanced machine learning technologies, to know how exactly lgorithms do what they do.

That's slowly starting to change, says Stewart.

Canadian organizations that process data of European Union (EU) residents will have to comply with the EU's General Data Protection Regulations (GDPR), which come into effect in May.

These new regulations place great emphasis on citizens' right to be given sufficient information about the automated systems used for processing their personal data.

"When I have a machine learning program that diagnoses cancer, recommends a specific treatment of chemotherapy or approves or doesn't approve a loan, I need to be able to say why it made that specific recommendation," he says. "There are now technologies that in some cases can interpret the way machine learning works."

Augmented reality will get down to business

Augmented reality is still often associated with cartoons and bunny ears, but Deloitte predicts it will make its way into instruction manuals, tech support services and marketing as the technology becomes increasingly photorealistic.

And while smartphone AR gets closer to worldwide adoption, it's not going to change the typical office environment where computers still reign supreme.

"AR is yet another feature on our all-powerful swiss army knife of smartphones and it's not going to replace existing office productivity technology like computers," he says.

But companies like Montreal-based Reacts, which has produced a platform for telehealth and other applications that supports multiple video streams, will likely enhance those functions with AR, says Anders McKenzie, the managing partner for the Technology, Media and Telecommunications (TMT) industry and senior partner in Deloitte's financial advisory practice.

"I have seen several of my clients in the medical space linking up facilities on a global basis ... they can quickly superimpose and even manipulate certain utensils and devices to help guide doctors in the field," he said.

More than a billion smartphone users will create AR content at least once in 2018, according to Deloitte.

Deloitte's Canadian TMT Predictions roadshow begins on Jan. 11 in Toronto. It will also make stops in Montreal, Quebec City, Vancouver, Ottawa and other Canadian cities.

Source: ITbusiness.ca