New techniques in machine learning will “unblock” new use cases, says Tim Ensor
Artificial intelligence is here, and can provide a serious competitive advantage, according to Tim Ensor, director of AI at Cambridge Consultants, who chaired the ‘Deliver’ stage at the recent AI Summit in San Francisco.
But challenges to wider adoption of new AI-powered
tools remain, chief among which is getting access to a large volume of data
necessary for algorithm training, and ensuring the high quality of data that’s
required for those algorithms to be accurate.
“We're seeing a whole host of new math
techniques coming through which are going to start allowing high-performing AI
systems to be able to be delivered for those applications where the data is
really difficult to come by,” Ensor told AI Business.
“And that's an area where we're researching
quite heavily. So, starting to bring some new techniques to the fore like
domain adaptation, few-shot learning, meta-learning, some of these newer
techniques, and starting to be able to get these into those applications where
today, they're kind of blocked. And I think by using these applications and
these areas of research, we're going to start unblocking those and see what
comes to the market.”
Cambridge Consultants won the Best Innovation in Deep Learning award at the AIconics event – part of the AI Summit celebrating the outstanding achievements of individuals, projects, teams and organizations that are responsible for harnessing breakthrough innovations in artificial intelligence.