Artificial Intelligence (AI) is arguably one of the most exciting technological advancements of recent years, with computers being able to process more than previously imagined and being utilised in countless fields, from military and aviation to finance and business.
Advancing the field of AI are C4DI members John Stamford and Brian Peach, PhD students at the University of Hull, who are researching ‘Machine learning techniques for clinical decision support’ and ‘Intelligent behaviour models for mobile robots’ respectively.
Together they founded NextAI, a research company exploring the latest developments in artificial intelligence and machine learning, with the aim of helping develop the skill sets of local companies to allow them to make use of these advancements in tech. They decided to bring accessible AI to the digital community in Hull after seeing a lot of investment start-up in UK-based research companies in areas like Cambridge and London, and in large tech-based cities overseas such as San Francisco.
A visualisation of how an AI system would process an image.
John explains how they work with existing AI to customise it for their needs: “We have started to explore the development of general AI systems which can be applied to a variety of contexts without changing the core architecture. As part of this, we have recreated Google Deepminds Deep Q-Network which learns to play Atari 2600 games without any external training.”
Their recent work has focused on the exploration and development of deep machine learning techniques which can be used with unstructured data such as images and text to simulate complex biological processes; imitating the cells of an eye for detecting the features of an image, and then neurons in the brain for learning and recognising any patterns.
A graph of the data collected and analysed by NextAI's system.
Brian explains how they shared their research, and how it resulted in becoming the first scientific research paper published by a C4DI-based company: “As a way of promoting the work we submitted a paper where it was peer reviewed by 3 leading academics. The paper was accepted and we received excellent feedback. We presented the work at the 2nd IET International Conference on Technologies for Active and Assisted Living (TechAAL 2016) in October and plan to release the image collection as a dataset so it can be used by other researchers.”
Congratulations to John and Brian, hopefully there will be plenty more successful papers from them!