![]() |
![]() |
Dr Vidhyalakshmi Karthikeyan (Head of Data and Insights, YouView TV Limited) ![]() Zero to ML
This talk is about the adoption of machine learning at YouView TV, the tech company behind the daily viewing experience of nearly 3 million consumers on BT & TalkTalk STBs and a number of Sony TVs. We will cover some of the key objectives and challenges in driving the adoption of ML in an organisation that is relatively early in its journey to unlock value from its data. We will progress to some use cases where we have successfully used ML within the organisation and outline our vision for these initiatives going forwards.
|
![]() |
![]() |
Dr. Laura Weis (Future of Work Lead, Satalia) ![]() AI-enabled Workforce Environments: empowering the new ‘net-work’
We are living through a fundamental transformation in the way we work, as we progressively move towards organisational models that are flatter, more fluid and collaborative, and increasingly dependent on knowledge assets. Work is increasingly becoming “net-work”, as the formal ‘command-and-control’ based hierarchies are replaced by more informal, dynamic networks that span across business functions and teams, incorporating a myriad of internal and external contributors. With this shift in structure comes a need to understand and intentionally lead it. This talk focuses on the critical role of advanced AI technology in the emergence and management of these interconnected workforce environments. While deep technical expertise is a must-have, success will also require a profound understanding of how to use this knowledge to drive increasingly complex business strategies. In this context, a people-centric approach has become increasingly important; an approach that continuously engages workers in this networked environment, fostering strategic and inclusive interaction and collaboration between people, people and work, and people and machines.
|
![]() |
![]() |
Dr Detlef Nauck (Principal for AI & Data Science, BT) ![]() What’s my AI doing?
Organisations are rushing to deploy AI solutions inspired by genuine business opportunities but sometimes also by hype or fear of missing out. Most of today’s AI solutions in organisations are new types of IT automation not driven by coded business logic but by statistical pattern recognition. This means we know that sometimes the AI will be wrong and that we need to mitigate risks in areas like bias/fairness, transparency, and accountability – to name a few. When organisations deploy vendor solutions instead of building them they face the additional challenge of finding out if the vendor follows some best practice guidelines for responsible AI that are compatible to their own. The advent of AI Regulation means that organisations not only need to convince themselves that their AI is under control, but that they will have to pass AI audits soon. In this talk I will take a look at the technical and organisational challenges we are facing when we are deploying and managing AI at scale.
|
![]() |
![]() |
Dr Sandy Brownlee (Senior Lecturer in Computing Science, Stirling University) ![]() Sustainable Building Design through Evolutionary Algorithms and Optimisation
In recent years, evolutionary algorithms have increasingly been applied to the optimisation of real-world industrial problems. Optimisation of building designs is one such area: typical designs have large numbers of variables, including construction materials, dimensions and equipment specifications. All of these can affect construction cost, operational energy use and occupant comfort. Given that the lifespan of a typical building is measured in decades, the environmental impact of getting the design right is large. The goal of Evolutionary Multi-objective Optimisation (EMO) is to find a set of designs representing a trade-off between conflicting objectives such as cost vs energy efficiency. This trade-off can be used to support designers in decision-making. I will explore a few approaches to supporting decision-making aimed at revealing what drives the trade-offs between energy consumption, costs, and comfort, for typical building design optimisation problems at the small scale (individual building) and large scale (region-level housing stock).
|
![]() |
![]() |
Dr. José Miguel Rojas Siles (Sheffield University) ![]() Search-based Automated Test Generation
Search algorithms are at the core of computer science, but searching for program inputs is not straightforward as the possible inputs for any real program is far too high to explicitly enumerate them all. Consequently, the search for effective tests needs to be informed by heuristics and needs to use algorithms that can cope with the complex structure and properties of programs under test and test data. Many different heuristics and algorithms have been proposed to address this problem. Search-based testing is a part of a larger domain in software engineering, where meta-heuristic search algorithms are used to solve complex software engineering problems. This talk presents a search-based approach to automatically generate executable unit tests for object-oriented programs and outlines current trends and challenges where modern AI can be applied in this context.
|
![]() |
![]() |
Professor Edward Keedwell (Professor of Artificial Intelligence, Exeter University) ![]() Augmented Evolutionary Intelligence: Using Humans and AI to Co-Design Solutions to Difficult Problems
The optimisation of the operation and design of systems is an important subfield of AI. When coupled with accurate simulations and digital twins, these methods can support the decision making process and even innovate new methods and designs through automated search and optimisation algorithms. However, these methods can lack an intuitive understanding of the system being optimised and including a human-in-the-loop has demonstrable benefits to the optimisation process. This talk will describe our work on augmented evolutionary intelligence, a method that beneficially combines domain experts, machine learning and evolutionary algorithms to co-create solutions to optimisation problems, whilst minimising user fatigue. The method has potential application to many domains and in this talk bin-packing and water distribution network optimisation will be presented as example case studies.
|