Machine learning algorithms developed by computer scientists from the University of Southampton have pinpointed a transfer target that would most suit Southampton FC in the January transfer window.

The AI insight tips that England Under-21s’ Brandon Williams would be an excellent fit for the Saints and offer a long-term replacement to 31-year-old left-back Ryan Bertrand.

The analysis suggests that Williams’ playing style and chemistry with his prospective teammates are ideally suited to the Southampton FC, with the club being much better for his development than other reported loan options such as Newcastle United FC.

Brandon Williams has moved down the pecking order at Manchester United this season, however it remains to be seen if the club will be willing to let the English prospect leave on loan.

Southampton FC have also been linked with Manchester City's Oleksandr Zinchenko, but the AI scout warns the left-back would not be as good a fit and may require higher wages and fees due to his age.

Sentient Sports, a new company launched by postgraduate researcher Ryan Beal, has devised a set of algorithms based on cutting-edge AI research that can optimise the decision-making and scouting process when buying and selling players. These are being utilised by AI Abacus, another leading start-up founded by Ramm Mylvaganam who is a business leader in football analytics.

They are already winning business from major football teams across Europe. They use AI to analyse the player’s chemistry (how he fits in with the new teammates) and style fit (how he fits the new teams tactics) to provide decision making support to top teams.

Ryan’s research in the Agents, Interaction and Complexity (AIC) Group is focussing on artificial intelligence in team sports. His new business venture continues his close collaboration with Gopal Ramchurn, Professor of Artificial Intelligence and Director of the Centre for Machine Intelligence, and former Southampton PhD student Tim Matthews.

"Professional team sport is a game of winners and losers and globally the stakes have never been higher," Ryan says. "Football, like other sports, has huge amounts of data associated with it, and some of this is already used by clubs to analyse the performance of their players. We are using our world-leading expertise to get more value out of that data and turn these insights into match-winning insights."

The Southampton experts are offering a teamwork algorithm that looks at how pairs and groups of players work together on the pitch to assess ‘player chemistry’; an algorithm that predicts the number of goals/assists a player will achieve; another that evaluates the suitability of a player in the style of play of a new team; and a final algorithm that looks at the cost benefit of players.

These algorithms are then used to predict how a player would perform if they were to play for a specified team.

"One of the key things our chemistry model shows for Saints is that the signing of Kyle Walker-Peters has been key this season,” Ryan says. “He has been shown to link up effectively with a number of the attacking players in the team and help boost the team overall."

Ryan is lifelong Southampton FC fan and is delighted with the club’s strong first half of the season which has placed the club in contention for European qualification. The club’s recent transfer dealings have all fitted well with Sentient Sports’ modelling.

"When Theo Walcott arrived on loan in the summer we looked at all the best options for him in the Premier League and Saints came out on top," Ryan explains. "We also identified Mohammed Salisu as one of the top young centre backs in Europe last season before he signed for Saints. So, I'm holding out hope that when he gets fit, he will be a very good long-term signing."

Read more about Sentient Sports in the latest edition of Re:action, the University's research and enterprise magazine. AI Abacus's work has been featured by the Sunday Times and the Daily Mail.

Electronic engineers at the University of Southampton are helping optimise the performance of a new wearable technology for the construction industry.

Energy harvesting experts in the Smart Electronic Materials and Systems research group have partnered with leading UK construction technology company Mafic to further development of its Safeguard Internet of Things (IoT) solution.

Safeguard is a wearable device that can fit into a hard hat, armband or pocket. Using machine learning, the devices can recognise the unique movement patterns of users completing different tasks and record exactly what is happening.

The new collaboration, funded by a grant from the £5 million SPRINT (SPace Research and Innovation Network for Technology) programme, will exploit technology developed within the School of Electronics and Computer Science.

Dr Alex Weddell, Lecturer in the SEMS group, says: “The key objectives of this SPRINT project are reducing the power consumption of the Safeguard device, harvesting energy and optimising the charging of the hard hat. We have a long history of working in energy harvesting, including the design and development of power management subsystems for CubeSats.

"More recently, we have explored wireless power transfer technology - which also has applications in space. We are excited to explore how energy harvesting from light, movement, or temperature differences can be used to extend the battery life of Mafic's Safeguard IoT device."

The project will allow Mafic customers to be less reliant on a power source and deploy Safeguard into remote environments with minimal supporting infrastructure such as on-board commercial ships, in offshore environments, or in 'not-spots' or remote construction sites.

Read the full story on the main news page.

Postgraduate researcher Iris Kramer from the University of Southampton is being supported by the Royal Academy of Engineering as she scales deep learning software that identifies buried ancient sites from space.

The archaeologist turned computer scientist has been awarded an Enterprise Fellowship for her ArchAI start-up, which is based on techniques developed in the Vision, Learning and Control research group.

The venture has also received backing from the UK Space Agency this month, through funding from the national Space Research and Innovation Network for Technology (SPRINT) programme.

Iris' AI solution helps smooth expensive planning processes for developers and saves historical sites from unnecessary destruction by automating archaeological assessments.

Iris says: "I'm delighted to receive this award and recognition from the Royal Academy of Engineering. Using ArchAI's technology over conventional techniques, developers could save hundreds of thousands of pounds in costs in addition to time savings of six months on a major housing or road development of 100 hectares.

"That's just one use for our technology and the Enterprise Fellowship programme will accelerate ArchAI towards addressing wide-ranging environmental challenges globally."

Space archaeology uses satellites or high-flying aircraft to take pictures of the Earth's surface to find hints of ancient features buried under the ground. Things may show up visually or near infrared may show small differences in vegetation, with growth on top of buried stone likely to be less healthy.

Dr Fraser Sturt, a Professor of Archaeology at the University of Southampton, says: "Aerial photography transformed archaeology in the early 20th century, revealing sites in a way that few people could have conceived of in the past. Advances in Earth Observation and Machine learning offer another leap forward, helping us to identify and monitor sites across of space and time. This information is critical not only for our understanding of the past, but how we manage the built environment and its development in the future."

Iris' PhD research is the first in the world to apply deep learning to the detection of archaeological sites from Earth Observation data. The project has trialled the deep learning techniques with Historic Environment Scotland to automatically identify hundreds of archaeological sites.

Last year, she became one of just six participants to be selected for the Ordnance Survey and HM Land Registry Geovation Accelerator Programme.

In October, she pitched for ArchAI in a Dragons' Den-style event hosted by the University's Future Worlds start-up accelerator. The business impressed the dragon investors and she received a £70,000 offer of investment at the highest ever valuation in the event's history.