Newcastle University is to spearhead a project to encourage ‘garage-level’ breakthroughs in artificial intelligence and machine learning.
Many of science’s greatest advances have resulted from people tinkering in their garages. The inventor of the telephone Alexander Graham Bell set up a workshop behind his father’s house, and Walt and Roy Disney experimented with making cartoons in a Hollywood garage.
In 1939, Hewlett-Packard built their first product in a garage at 367 Addison Avenue, Palo Alto – now known as the ‘birthplace of Silicon Valley’.
Now the Machine Learning Garage project is helping north-east tech start-ups access the support they need to enhance the UK’s reputation for AI innovation.
Machine Learning Garage will draw on the expertise of academics from Newcastle University’s National Innovation Centre for Data, as well as the skills of big tech players such as Amazon Web Services, Google Cloud Platform and Graphcore.
In addition to expertise, the project aims to give tech start-ups access to computational power. It is hoped that this will help break down the barriers that hamper new businesses in this field.
A study by the organisation Digital Catapult – which surveyed 10% of the UK’s AI and machine-learning start-ups – found that more than half felt constrained by their lack of access to computational power.
The Machine Learning Garage project will be led by Dr Steve McGough, a senior lecturer at Newcastle University’s School of Computing.
Dr McGough said, “The UK is already world-leading in the areas of machine learning. However, we are yet to see this great research trickle out from academia and into the economy.”
“This is a push to make this happen.”
“Machine learning is any situation in which a computer ‘learns’ to solve a problem by reacting to inputs rather than being programmed through instructions telling it what to do.”
“Here at Newcastle we take this one step further – using an approach known as Deep Learning.”
“This is where artificial ‘neurons’ are connected together to ‘mimic’ the processes in the brain. By connecting many of these neurons together in deep networks, it is possible to train a computer to solve many complicated tasks.”
“These have included the ability to recognise what is in a picture, performing speech to text conversion and even winning at the game Go.”
Dr Marko Balabanovic, the chief technology officer at Digital Catapult (the provider of the Machine Learning Garage project), said, “Machine Learning Garage is part of a broad programme at Digital Catapult to accelerate the adoption of machine learning technologies across UK companies and industries, especially in areas like digital manufacturing or creative industries where there are potential large gains in growth and productivity.”
The Machine Learning Garage project is open to all companies engaged with developments in artificial intelligence and machine learning.
The open call went live on 23rd January, with applications to roll every six weeks thereafter.
The project will run for three years in total. For more information, please go to www.migarage.ai.
(Featured image courtesy of Matthew Hurst, from Flickr Creative Commons)